Clinical and molecular characteristics of Klebsiella pneumoniae ventilator-associated pneumonia in mainland China

Background Klebsiella pneumoniae is a prominent nosocomial pathogen that accounts for up to 10 % of all hospital-acquired infections. It is a frequent cause of ventilator-associated pneumonia (VAP). The purpose of this study was to investigate the clinical characteristics of K. pneumoniae-associated VAP and the molecular characteristics of K. pneumoniae strains. Methods We retrospectively reviewed 70 mechanically ventilated patients with K. pneumoniae isolated. All K. pneumoniae strains were examined to determine hypermucoviscosity (HV) phenotype, capsular serotypes, virulence genes, multilocus sequence typing and antimicrobial susceptibility. Results Hypermucoviscosity was found in 14 of 70 (20 %) isolates of K. pneumoniae. Among the 70 patients, 43 cases (61.4 %) developed VAP. Furthermore, VAP was more frequently induced by HV-positive K. pneumoniae (14/14, 100 %) than by HV-negative strains (29/56, 51.7 %). HV-positive K. pneumoniae-associated VAP patients were more inclined to develop bacteremia and had a higher mortality rate than HV-negative strains VAP patients. Antibiotic resistance was more frequent in HV- negative strains- than in HV- positive strains-infected patients. The prevalence of rmpA and aerobactin genes were 85.7 % and 85.7 % respectively, and serotypes K1 and K2 accounted for 14.3 % and 28.6 % of the hypermucoviscosity strains, respectively. Strains carrying rmpA and aerobactin genes were significantly associated with HV-phenotype, and rmpA and aerobactin coexisted in HV-positive strains. Multilocus sequence typing analysis identified 24 different sequence types from K. pneumoniae VAP samples. Conclusions HV-phenotype is the major virulence determinant for mechanically ventilated patients. There was a specific sequence typing (ST) distribution between HV-positive and HV-negative strains.

It has been reported that the HV-positive phenotype, certain serotypes, and the presence of rmpA and aerobactin genes are virulence determinants in K. pneumoniae infection [13][14][15][16][17]. However, there have been few reports on the specific roles of these factors in VAP in mainland China. In the present study, we used isolates collected from mechanically ventilated patients to delineate the clinical characteristics of K. pneumoniae-associated VAP and molecular characteristics of K. pneumoniae strains observed over a 2-year period.

Hospital setting and study population
The Henan Provincial People's Hospital is a 3900 bed tertiary care hospital with 6 ICU wards with an approximate annual admission of 1600 ICU inpatients. K. pneumoniae strains were collected via endotracheal aspiration from mechanically ventilated patients with suspected pneumonia and stored at −80°C before use. Medical records of patients from whom the collected strains were isolated were reviewed between January 2012 and August 2014. VAP was diagnosed in these patients who fulfilled both the clinical and microbiological criteria. The clinical criteria for the diagnosis of VAP are the presence of a new pulmonary infiltration on chest radiography plus at least two of the following: fever above 38°C, purulent secretions, and leukocytosis or leucopenia [18]. The microbiological criteria are quantitative tracheal aspirate culture with ≥10 5 CFU/mL and positive gram stain (>10 polymorphonuclear cells/lowpower field and ≥1 bacteria/oil immersion field with or without intracellular bacteria) [19][20][21]. The VAP diagnosis was reconfirmed by two infectious diseases specialists independently. Polymicrobial infections were excluded from the analysis.

Data collection and microbiologic analysis
The following clinical information was collected for each patient: demographic characteristics, VAP diagnosis, reasons for mechanical ventilation, laboratory data, and chest radiograph reports. Bacteremic VAP was diagnosed when blood and respiratory samples yielded the same microorganism and other sources of infection that could account for the bacteremia were absent. Furthermore, blood and respiratory cultures were performed within 48 h.
Outcome was defined as in-hospital mortality measured 30 days after the onset of VAP. Trauma was defined as the presence of injury in more than one body area or system or the presence of major cranial trauma alone. Chronic lung diseases included bronchiectasis, chronic obstructive pulmonary disease. The initial laboratory value was defined as that measured within 48 h of the onset of VAP. The BD Phoenix system (Becton Dickinson, USA) was used to confirm bacterial identification. Antibiotic susceptibility was tested with the disk diffusion method, and interpretations were made according to the guidelines of the Clinical and Laboratory Standards Institute [22]. All of the K. pneumoniae isolates were screened and confirmed by a double-disk synergy test for produced extended-spectrum β-lactamase (ESBL).

Detection of HV-phenotype
For HV-phenotype determination, a standard bacteriologic loop was used to stretch a mucoviscous string vertically from a colony. The formation of a viscous string of >5 mm confirmed the HV-positive phenotype.
Serotyping, rmpA and aerobactin gene detection with PCR PCR was performed to amplify genes specific for serotypes K1/K2 and the rmpA and aerobactin genes as described previously [17,23]. K. pneumoniae ATCC9997 (K2) was used as a control strain. A bacterial colony from an overnight culture was added to 500 μL water and boiled for 15 min to release the DNA template. PCR was performed with the following conditions: 95°C initial denaturation for 5 min followed by 30 cycles at 95°C for 30 s, 55°C for 30 s, 72°C for 90 s, and a final extension at 72°C for 5 min.
Statistical analysis SPSS 17.0 was used for statistical analysis. The chisquare or Fisher's exact test was used to analyze contingency data, and continuous data were analyzed with the Student's t test. A P value of <0.05 was considered significant, and all probabilities were two-tailed.

Antimicrobial susceptibility test
The prevalence of HV-negative isolates exhibiting resistance to the tested antimicrobials was higher than that of the HV-positive isolates ( Table 4). The detection rates of ESBL-producing K. pneumoniae isolates were 46.5 % (20/43). The percentage of ESBL-producing HV-negative isolates was significantly higher than that of ESBL-producing HV-positive isolates (62.1 % compared to 14.3 %, P < 0.05). Among all of the isolates, one of them (1/43, 2.3 %) was resistant to imipenem and meropenem.   analysis showed that all of these isolates belonged to two major MLST groups, ST1419-like and ST515-like (Fig. 1b). ST11 (n = 7) and ST15 (n = 7) were the prevalent STs in HV-negative isolates.

Discussion
In the retrospective study, we analyzed 70 isolates of K. pneumoniae from mechanically ventilated patients between January 2012 and August 2014. We observed that HV-positive strains accounted for 20 % (14/70) K.
pneumoniae isolates. The genotypes of K1, K2, rmpA and aerobactin were only positive for HV-positive strains. The medical records showed that 43 strains accounted for 61.4 % (43/70) induced VAP in mechanically ventilated patients. There was difference in ST distribution between HV-positive and HV-negative stains in VAP patients. We further observed that the occurrence rate of VAP was significantly higher for HV-positive strains infection than HV-negative strains infection (100 % vs 51.7 %, P < 0.05). Furthermore, the results showed that bacteremic VAP occurred more frequently in HV-positive strains than in HV-negative strains (35.7 % vs 3.4 %, respectively, P = 0.017). The mortality rate was higher in the HV-positive group (57.1 %) than in the HV-negative group (13.8 %). Laboratory data also showed that HV-positive strains VAP patients had increased frequency of higher C-reactive protein (CRP) and lower albumin levels. CRP is an acute-phase protein that has been evaluated in the critical care setting [24]. Moreover, CRP values correlated well with the severity of the infection. Hillas et al. [25] observed that a rise in CRP between days 1 and 7 increases the risk of septic shock. Albumin level reportedly decreases as part of negative acute-phase protein. Previous studies have indicated that the degree of hypoalbuminemia in critically ill patients correlates with the intensity of the inflammatory response caused by infection [26,27]. These findings indicated that compared with HV-negative strains, HV-positive strains are more virulent in the development of VAP.
Capsular serotypes, especially serotypes K1 and K2, are important virulence factors for K. pneumoniae [28,29]. Yu et al. reported that the capsular serotypes K1 and K2 have particular virulence and were more common in patients with community-acquired pneumonia (23/49 isolates, 47 %) than in those with hospital-acquired pneumonia (2/18 isolates, 11 %) [30]. Our results were similar to the study. We also found that capsular serotypes K1 and K2 comprised only 4.7 % (2/43) and 9.3 % (4/43), respectively, of all K. pneumoniae isolated from VAP patients. In addition, the HV-positive strains accounted for 32.6 % (14/43) K. pneumoniae in VAP patients. These results suggest that hypermucoviscosity rather than serotype K1 or K2 has become a common pathogen of VAP in mechanically ventilated patients.
Previous study has shown that K1 serotype and rmpA are associated with HV-phenotype in K. pneumoniae [17]. In this study, we found that HV-phenotype frequently coexists with the rmpA gene. In addition, the prevalence of the K1 serotype in HV-positive strains was only 14.3 % (2/14). Chen et al. further confirmed that the rmpA gene is a regulator of HV-phenotype [31]. Loss of this gene leads to the loss of HV-phenotype. These data showed that the rmpA gene rather than the K1 serotype correlated with HV-phenotype in K. pneumoniae. Meanwhile, the results also showed that two HV-positive strains which did not possessed rmpA gene. This indicated that there may be other regulatory mechanisms for expression of HV-phenotype. In addition, we found that rmpA gene coexists with the aerobactin gene in HV-positive K. pneumoniae. This result was in line with the observations of Yu et al. [30]. Further investigation showed that rmpA is located on a 180-kb virulence plasmid, which also contains many virulence-associated genes, including aerobactin for iron acquisition [15,32]. However, we did not determine which of the two virulence factors (rmpA or aerobactin) of HV-positive strains is more critical in VAP.
Among these VAP patients, the HV-positive isolates were significantly more susceptible to the antimicrobial agents that were tested, when compared with the HV-negative isolates. However, we found two ESBLproducing HV-positive isolates. Previous study also showed that most HV-positive strains were very susceptible to antimicrobials [33]. Nonetheless, some cases of infection due to multidrug resistant HV-positive K. pneumoniae have already been described [34]. Therefore, management of VAP due to HV-positive isolates will become extremely challenging.
In the present study, 24 STs were observed in the 43 K. pneumoniae isolates from VAP patients. Eleven MLSTs were identified in 14 HV-positive strains, suggesting a polyclonal origin. Thirteen STs were observed in 29 HV-negative strains, among which ST11 (n = 7) and ST15 (n = 7) were the most prevalent. We noticed that a specific ST distribution occurred between HV-positive and HVnegative strains, suggesting the difference genetic background existed among the 2 phenotype isolates. This study had several limitations. First, it was a retrospective study from a single hospital and the small sample size may have had selection bias. Second, the diagnosis of VAP in mechanically ventilated patients is difficult, and still there is no "gold-standard" diagnostic method. Third, the pathogenic mechanism of HV-positive K. pneumoniae in VAP is unclear and requires further investigation in future studies.

Conclusion
We have shown that HV-phenotype, rather than serotype K1 or K2, was the major virulence determinant for mechanically ventilated patients. Patients infected with HVpositive strains were more likely to develop VAP and bacteremic VAP. Furthermore, HV-positive K. pneumoniae VAP had a higher mortality than HV-negative strains VAP. We hope our results will draw attention from physicians, which may lead to prompt recognition and successful management of HV-positive K. pneumoniae VAP.