CRKP strains and Virulence-associated features
A total of 80 CRKP strains isolated from 80 patients in ICU were included in the study. The percentages of 80 CRKP strains isolated from respiratory tract, urine, blood, pus and puncture fluid were 72.5%, 15.0%, 6.3%, 5.0% and 1.3%, respectively. CRKP strains were divided as CR-hvKP and carbapenem-resistant non-hypervirulent Klebsiella pneumoniae (CR-non-hvKP) based on whether they contained virulence genes. The detection of the presence of virulence genes showed that 51 (63.8%) of 80 CRKP strains were positive, suggesting carrying virulence plasmids, which were designated as CR-hvKP group, while the remaining 29 strains were designated as CR-non-hvKP group.
Among the five virulence-associated genes, rmpA2, iucA, iroN, peg-344 and rmpA harboured by 51 strains (Figure 1), the most common virulence gene was rmpA2, the prevalence was 90.2% (46/51), followed by iucA, iroN, peg-344 and rmpA, the detection rate were 88.2% (45/51),27.5% (14/51), 21.6% (11/51) and 19.6% (10/51), respectively. Seven (13.7%) strains had all five of the virulence genes, which indicate that the full length of the virulence plasmid pLVPK could be harbored by these strains. The results of the string test showed that 41.2% (21/51) CR-hvKP strains showed negative results in string test, whereas the results of the 17 CR-non-hvKP strains were positive, accounting for 58.6% (17/29).
For the eight serotypes closely related to hvKP, 65(81.3%) strains were positive for K47, K54, or K64 serotypes, additionally, there is no strain with K1, K2, K5, K20 or K57 serotypes in the present study. Serotype K64 was dominated serotype of CRKP strains [62 (77.5%) of 80 strains]. The frequency of K64 type (86.3% vs. 62.1%, P=0.013) identified in CR-hvKP strains was significantly higher than that of CR-non-hvKP strains. K-non-typable (34.5% vs. 9.8%, P=0.007) was strongly associated with CR-non-hvKP. Only two strains were identified as K47 or K54 (Table 1).
Susceptibility results and Carbapenemase resistance Genes
The antimicrobial susceptibility profiles of the CR-hvKP and CR-non-hvKP strains were listed in Table2. According to the AST results, all strains were resistant to imipenem and meropenem with exception of one CR-hvKP strain being susceptible to memropenem, and showed high level resistance to cephalosporins, monobactam, β-lactams/βlactamase inhibitor combinations, fluoroquinolones, but were almost not resistant to ceftazidime/avibactam (2.0% vs. 3.5%), imipenem/avibactam (3.9% vs. 3.5%), polymyxin B (0.0% vs. 17.2%) and tigecycline (0.0% vs. 0.0%). Ceftazidime and imipenem by combining avibactam (4 µg/mL) significantly decreased the MIC90 values more than sixteen-fold than that of ceftazidime and imipenem alone against KPC-2-producing Klebsiella pneumoniae. Although we found there were no statistical significance in the resistance rates of antimicrobial agents between the two subgroups (p ≥ 0.05), except polymyxin B (P=0.002). The MIC50 values of the following antibiotics of the CR- hvKP strains increased two-fold compared with those of CR-non-hvKP: ceftazidime (>128 vs. 128μg/mL), ceftazidime/avibactam（4/4 vs. 2/4 μg/mL), imipenem/avibactam (0.5/4 vs. 0.25/4 μg/mL), tigecycline (2 vs. 1μg/mL), levofloxacin (64 vs. 32μg/mL), cefoperazone/sulbactam (>256/128 vs. 256/128μg/mL).
According to phenotypic experiment results and detection of carbapenemase genes, all CR-hvKP and 96.6% (28/29) CR-non-hvKP strains showed only produced blaKPC-2carbapenemase gene and the presence of KPC -lactamas. Only one CR-non-hvKP strains showed the presence of MBL enzyme, harbored blaIMP-4.
MLST analysis revealed that ST11 and ST15 were the most common STs in CRKP strains, and ST198 and ST438 were distributed in one strain of CR-hvKP strain and one strain of CR-non-hvKP strain. The frequency of ST11 type (88.2% vs. 69.0%, P=0.034) identified in CR-hvKP strains was significantly higher than that of CR-non-hvKP strains, however, the proportion of ST15(9.8% vs.27.6%, P=0.039) strains was lower (Table 1). The strains were divided into several clones by combination of capsular genotypes and STs. We found that ST11-K64 CRKP strains (86.3% vs.58.6%, P=0.005) were strongly associated with CR-hvKP, while ST15- K-non-typable (7.8%vs. 27.6%, P=0.017) strains were correlated with CR-non-hvKP (Table 1).
Clinical factors for CR-hvKP infection of elderly patients
Demographic and clinical factors of patients with CR-hvKP and CR-non-hvKP infections are summarized in Table 1. The age, gender, antibiotic exposure, hospitalization within the last 90 days, or length of stay from CRKP isolation to outcome (In-hospital mortality or discharge) didn’t have significant difference between CR-hvKP and CR-non-hvKP (P>0.05). In addition, there was no statistical difference among most underlying conditions such as pulmonary disease, diabetes mellitus, hypertension, cerebrovascular diseases, malignancy, hypoproteinemia and liver abscess (P>0.05), except cardiovascular disease (60.8% vs.17.2%, P<0.001). Gastric tube was more prevalent in CR-non-hvKP group than in CR-hvKP group (96.6% vs.80.4%, P=0.045), whereas the other invasive procedures and devices had no significant difference. Most of the outcomes between the two groups (P>0.05) had no statistical difference, the frequency of change of initial antibiotics due to clinical worsening was higher in CR-hvKP group than CR-non-hvKP group (58.8% vs.34.5%, P=0.036) (Table 1).
Risk factors of CR-hvKP infection for the elderly patients
Nine variables with P<0.05 in univariate analysis were included in the multivariate analysis, including K64, K-non-typable, ST11, ST15, ST11-K64, ST15-K-non-typable, cardiovascular disease, gastric tube, change of initial antibiotics due to clinical worsening. The results showed that ST11-K64(odds ratio [OR] =8.385; P=0.003) and cardiovascular disease (OR=11.956; P＜0.001) were independent predictors of CR-hvKP infections for elderly patients.(Table 3).