Clinical and microbiological characteristics of Klebsiella pneumoniae liver abscess in East China

Background Klebsiella pneumoniae has been the dominant pathogen for liver abscesses in several Asian countries. Although the prevalence of K. pneumoniae liver abscess (KLA) in mainland China is increasing recently, the clinical and microbiological characteristics of KLA in China have not been elucidated. Methods Clinical and microbiology characteristics of 45 consecutive patients with KLA from a tertiary teaching hospital in China between June 2008 and June 2012 were retrospectively evaluated. Results Vast majority of the strains were susceptible to main antimicrobial agents. Most of K. pneumoniae strains from pyogenic liver abscess patients belonged to K1/K2 serotype (68.9% for K1 serotype and 20% for K2 serotype). All K. pneumoniae strains were rmpA positive, and 68.9% of these strains were magA positive. Overall, 57.8% (26/45) of K. pneumoniae strains belonged to ST23. Twenty-five of 26 ST23 K. pneumoniae isolates (96.2%) from KLA patients were magA-positive and K1 serotype. Only 28.9% (13/45) of KLA isolates exhibited hypermucoviscous phenotype, which is clinically used as the characteristic of hypervirulent K. pneumoniae (hvKP). Liver abscess sizes in patients infected with hvKP were tend to be larger than those in patients infected with cKP. There was no significant association between the microbiological and clinical characteristics including serotypes, magA and rmpA genotypes, and STs with the metastatic infection and prognosis of KLA. Conclusions Neither the serotypes, magA and rmpA genotypes, nor the STs of K. pneumoniae were associated with the metastatic infection and prognosis of KLA. However, further studies with larger sample are needed in the future. Electronic supplementary material The online version of this article (doi:10.1186/s12879-015-0899-7) contains supplementary material, which is available to authorized users.


Background
Klebsiella pneumoniae has emerged as the dominant cause of pyogenic liver abscess in Asia and then was found worldwide. This condition is frequently associated with severe complications, including septic endophthalmitis and other extrahepatic lesions infections, especially in patients with diabetes [1][2][3]. This new K. pneumoniae variant was defined as hypervirulent K. pneumonia (hvKP). Furthermore, the ability for metastatic spread of infection demonstrated by hvKP is different from classic K. pneumoniae (cKP). Given there is no clear marker for hvKP strains, a hypermucoviscous phenotype has been considered to be associated with strains that cause KLA. This mucoid phenotype might be indicative of the extent of capsular polysaccharide expression, which is related to resistance to phagocytosis. Currently, hepato-virulent K. pneumoniae causing primary hepatic abscesses has posed a challenge for early laboratory identification and recognition. Bacteraemic K. pneumoniae isolates with positive string tests should be considered invasive strains capable of causing disseminated infection because the increased virulence associated with hyper virulent (HV) strains [4,5].
Capsular serotypes, magA, and rmpA have been documented in high prevalence for K. pneumonia liver abscess. Several studies of bacterial pathogenesis in Taiwan have documented that serotype K1 or K2, magA, and rmpA are possible virulence factors in K. pneumoniae liver abscess. The magA gene actually corresponds to the capsular polysaccharide synthesis (cps) gene wzy of K. pneumoniae isolates of serotype K1 [6][7][8][9][10]. The rmpA gene is a plasmid-mediated regulator of extracellular polysaccharide synthesis, and rmpA-carrying strains were associated with the hypermucoviscosity phenotype [8,11,12].
Multilocus Sequence typing (MLST) is an increasingly common used molecular epidemiologic approach for categorizing strains, and a typing screen has been described for K. pneumoniae. ST23 was most commonly described so far and is strongly associated with the K1 capsular serotype and liver abscess [2,13]. However, ST-65-like and −86-like are the two major MLST types among serotype K2 isolates from Asia [2,14].
In mainland China, the prevalence of K. pneumoniae liver abscess (KLA) is high, while only some KLA cases were reported in the English language scientific literature. Recently, Luo et al. firstly reported the molecular epidemiology and prevalence of virulence factors of the 51 KLA isolates in a Chinese hospital. However, very little clinical information was collected and evaluated in their study [15]. In our study, we analyzed clinical and microbiological characteristics of 45 KLA from a medical center in China between June 2008 and June 2012. To investigate the relationship of the clinical and microbiology characteristics of KLA, the present study evaluated the clinical manifestations of patients and detected the antimicrobial agents susceptibility, K1/K2 serotypes, MLST, hypermucoviscous phenotype, and magA/rmpA genotypes of the related K. pneumoniae isolates from KLA patients in a medical center over a 4-year period in East China.

Study design and setting
The study was a retrospective review of all patients with KLA in the First Affiliated Hospital, College of Medicine, Zhejiang University, an urban, 2500-bed major tertiary teaching hospital in Hangzhou, East China, from June 1, 2008 to June 30, 2012. The institutional review board of the First Affiliated Hospital, College of Medicine, Zhejiang University approved the study protocol and waived from the need for a consent form.
The hospital was one of the leading hospitals for liver transplantation in China and has an approximate annual admission of 108,700. It has a quaternary referral hepato-biliary unit and therefore, many patients with hepato-biliary diseases were transferred from rural hospitals to our hospital. In 2013, it became the one of the public hospitals in mainland China to be accredited by the Joint Commission International, a US-based, World Health Organization-authorized organization for medical quality evaluation.

Data sources and definition of KLA and outcomes
The medical records were retrospectively reviewed to extract all the patients with a diagnosis of KLA during the study period. A diagnosis of a KLA was defined with the combination of presence of the typical clinical manifestations of infection, such as fever, sepsis and right upper abdominal pain, imaging evidence, and positive aspiration that was consistent with a pyogenic liver abscess (PLA). The patients with KLA who were included in our study met the following criteria: 1) a PLA was the primary cause of the hospitalization but not a complication; 2) older than 18 years old; 3) PLA patients with K. pneumoniae culture-positive of drained abscess. Thus, PLA patients without drainage but with positive blood cultures of K. pneumoniae were excluded in this study. As practiced in our hospital, lab examinations for all patients with KLA are required on the admission and generally every 2-3 days at the discretion of the attending physicians.
For each enrolled patient, the following data elements were extracted: 1) demographic characteristics (age, gender); (2) coexisting conditions; (3) location and size of abscess; (4) laboratory examinations; (5) hospital outcomes. Primary outcome was hospital mortality, and secondary outcome were complications including septic shock and metastatic infections to extra-hepatic sites, such as spontaneous bacterial peritonitis, pneumonia, endophthalmitis.

Extended-spectrum beta-lactamases (ESBLs) detection
The combination-disk synergy tests using cefotaxime (30 μg) ± clavulanic acid (10 μg) and ceftazidime (30 μg) ± clavulanic acid (10 μg) were performed to detect ESBLs phenotype for all the collected isolates. The ESBLs phenotype was confirmed by 5-mm or greatly increased zone diameter for either cefotaxime or ceftazidime in combination with clavulanic acid versus its zone when tested alone.

String test
Strains with the hypermucoviscosity phenotype were defined as high virulent. A string test was performed to distinguish hvKP from cKP. A positive string test was defined as the formation of a viscous string >5 mm in length when bacterial colonies on an agar plate are stretched by an inoculation loop [5].
PCR-mediated detection of rmpA and capsular serotype-specific genes Genomic DNA was extracted from all K. pneumoniae strains (QIAGEN DNA extraction kit, QIAGEN, USA) and the rmpA and serotype-specific genes for the K1, and K2 capsular serotypes were amplified by polymerase chain reaction (PCR) as previously reported [16,17].
Total DNA was prepared and PFGE was performed as described previously [18]. The PFGE dendrogram was created by Software BioNumerics 6.6. The results were interpreted according to the criteria suggested by Tenoval et al. [19].

Statistical analysis
Statistical analysis was performed using SPSS 16.0 (Chicago, Ill, USA). Descriptive data were reported as either mean ± SD or number and percentage. Normally distribution of the data was tested with Kolmogorov-Smirnov test. With respect to the differences in outcomes between various groups, categorical variables were compared using chi-square analysis. Where the number of cases was smaller than 5, the Fisher's exact test was used. Continuous variables were compared using Independent Sample T test for normally distributed data and Mann-Whitney U test for non-normally distributed data. Significance was defined as a P value <0.05.

Patients' characteristics
From June 2008 to June 2012, a total of 45 liver abscess patients admitted to our hospital were diagnosed as KLA with the K. pneumoniae culture-positive of their liver abscess drainage. Demographic data and clinical characteristics were shown in Table 1.
Microbiologic characterization of the K. pneumoniae strains Only 1 of 45 K. pneumoniae strains was ESBL-positive. All the isolates were resistant to ampicillin, but eleven were unsusceptible to piperacillin, two to cefuroxime, one to cefotaxime, one to ciprofloxacin, one to gentamicin, and one to cefepime ( Table 2). All the isolates were susceptible to amikacin, amoxicillin-clavulanate, aztreonam, ceftazidime, cefoperazone-sulbactam, SMZ-TMP, imipenem, and meropenem, as shown in Table 2.
Virulence genes analysis showed 31 strains (68.9%) were magA gene positive and all 45 strains were rmpA gene positive, as shown in Table 2. According to the string test results, 13 hvKP isolates were obtained at the rate of 28.9%. All of the 13 hvKP strains belonged to K1/K2 serotype (7 K1 isolates and 6 K2 isolates).
Comparison of microbiological characteristics between ST23 and non-ST23 K. pneumoniae isolates As shown in Table 3, the prevalence of K1 serotype or magA-positive isolates in ST23 K. pneumoniae isolates was significantly higher than that in non-ST23 isolates (96.2% vs 31.6%, P < 0.001). Conversely, the prevalence of K2 serotype in ST23 K. pneumoniae isolates was significantly lower than that in non-ST23 isolates (0% vs 47.4%, P < 0.001). Table 4, the prevalence of K2 serotype in hvKP isolates was significantly higher than that in cKP isolates (46.2% vs 9.4%, P = 0.011). All of the 5 non-K1/K2 serotype isolates belonged to cKP, including 1 ST23 and 4 non-ST23 isolates. No significant difference was found between the magA positive rates in hvKP and cKP strains (53.8% vs 75%).

As shown in
Comparison of clinical characteristics of KLA patients infected with different K. pneumoniae isolates Comparison of clinical characteristics between KLA patients infected with magA-negative and -positive K. pneumoniae isolates As shown in Table 5, there is a trend that more patients with biliary tract diseases in KLA patients infected with magA-negative K. pneumoniae isolates than those in patients with magA-positive K. pneumoniae isolates (57.1% vs 25.8%, P = 0.053). There were no significant differences of liver abscess sizes and complications between magApositive and -negative K. pneumoniae isolates.

Comparison of clinical characteristics between KLA patients infected with hvKP and cKP isolates
More patients infected with cKP tend to have a history of diabetes mellitus or drinking than that of hvKP patients (56.3% vs 30.8% and 37.5% vs 15.4% respectively,   . There were no significant differences of liver abscess sizes and complications between ST23 and non-ST23 K. pneumoniae isolates (Table 5).

Discussion
Pyogenic liver abscess is a potentially life-threatening disease. Although Escherichia coli was the most common pathogen in liver abscess before the 1980s, K. pneumoniae had become the dominant pathogen for liver abscess during the past two decades. In Taiwan, over 80% of bacterial liver abscesses were caused by K. pneumoniae. In Singapore and South Korea, the prevalence of KLA was similar. In our medical center, approximate 81.7% of pyogenic liver abscess were caused by K. pneumoniae. Male gender, patients with diabetes mellitus and biliary disease were more likely to have KLA. Most patients had a single abscess on the right hepatic lobe. The low mortality rate has been similarly observed in recent reports on liver abscess [20,21]. Although liver abscesses caused by extended spectrum β-lactamase (ESBL)-producing K. pneumoniae have been reported, it is a rare occurrence [21]. In our study, only 1 K. pneumoniae strain produced ESBL, which was isolated from a KLA patient who had an accident injury one month ago. As previous reported, K. pneumoniae isolates from KLA were susceptible to almost all kinds of antimicrobial agents such as third generation cephalosporins, β-lactamase inhibitor compounds, and carbapenems [2,22].
The K1 K. pneumoniae isolates contributed to 68.9% of KLA in this study, which was higher than previous reports from mainland China (43% and 39.2%) [15,20,23]. The prevalence of K1 isolates was significantly higher than K2 in this study (68.9% vs 20%). Our MLST data of the K. pneumoniae isolates from KLA patients revealed that ST23 was predominant sequence type with a rate of 57.8%, which was higher than that in the recent report from North China (37.2%) [15]. This indicated that serotypes and MLST of KLA K. pneumoniae maybe different among various regions in China. Nationwide study would be necessary to realize the national molecular epidemiology of KLA K. pneumoniae in China. In our study, serotype K1 KLA K. pneumonia isolates that mainly belong to ST23, while ST65-like and ST86-like  [14]. PFGE results demonstrated that there was no clonal dissemination among the 26 ST23 K. pneumoniae isolates. Most of ST23 K. pneumoniae isolates (96.2%) belonged to K1 serotype with magA positive. As previously reported, ST23 was the most prevalent sequence type among serotype K1 K. pneumoniae isolates from both liver abscess and stool samples in the Asia Pacific region [2,[24][25][26]. It implies that liver abscess might develop after leakage of K. pneumoniae from a patient's bowel into their liver via the portal circulation [25][26][27].
magA has been described as the causative gene for K. pneumoniae liver abscess and septic metastatic complications [7]. The enzyme encoded by magA, also named wzy in accordance with the bacterial polysaccharide gene nomenclature scheme, functions as a polymerase involved in capsule synthesis, and this function is restricted to the capsular gene cluster of serotype K1 only [10,28]. In previous study, magA has been reported in 98.1% and 83.3% of K. pneumoniae strains isolated from patients with liver abscess and was significantly more prevalent than the bacteremic strains [17,28,29]. However, our study showed that magA was detected in 68.9% of K. pneumoniae isolates from KLA patients, and 96.2% in ST23 K. pneumoniae isolates from KLA patients. Therefore, almost all ST23 K. pneumoniae isolates from KLA patients was magA-positive and K1 serotype, but not all of the magA-positive and K1 serotype K. pneumoniae isolates from KLA patients belonged to ST23.
rmpA has been confirmed to regulate capsular polysaccharide synthesis and was proposed as a virulent factor in addition to magA and capsular serotypes K1/K2 [8,30]. Recently, rmpA-associated hypermucoviscosity phenotype has also been reported to play an important role in invasive purulent diseases caused by K. pneumoniae. K. pneumoniae serotypes K1 and K2 isolated from patients with liver abscess usually carrying hypermucoviscosity [2,5]. Our study showed that all K. pneumoniae strains that cause liver abscesses were rmpA-positive. However, our string test results revealed that only 28.9% K. pneumoniae isolates from KLA showed hypermucoviscosity. The hvKP rate was significantly lower than that reported by previous studies (more than 85%) and also the recent report from mainland China (70.6%) [5,8,15]. We had repeated the string test for three times and received the same results. It seemed that string test might not be a reliable method to identify K. pneumoniae with the potential to cause KLA. All of the 13 K. pneumoniae strains with hypermucoviscosity (hvKP) belonged to K1/ K2 serotype. The prevalence of cKP in K. pneumoniae isolates from diabetes mellitus and drinker patients were higher than that of hvKP. It was consistence with that hvKP was more liable to cause KLA in healthy hosts than cKP [5]. However, there was no significant difference of complications and prognosis between patients infected with hvKP and cKP, ST23 and non-ST23, or magA-positive and -negative strains.
Our study also has several limitations. First, any potential for selection bias in terms of those PLAs aspirated or those patients presenting to this particular institute. Second, that there is no denominator data here; neither in terms of number of PLAs in total, not of number of K. pneumoniae invasive isolates (i.e. positive blood cultures) which did not result in PLA. Third, although there was no significant difference of complications and prognosis among patients infected with different K. pneumoniae isolates, the sample size of our study was relative small. In order to determine the predictor of metastatic infection and prognosis of KLA, more KLA cases should be collected for further study.

Conclusions
In summary, K1 serotype and ST23 were the predominant serotype and sequence type of KLA K. pneumoniae in our study. The hvKP rate was significantly lower than that in previous reports. There was no significant association between the microbiological characteristics including serotypes, magA and rmpA genotypes, and STs with the metastatic infection and prognosis of KLA in this study. Further studies with larger sample are needed in the future.

Additional file
Additional file 1: Table S1. MLST and serotype analysis in K. pneumoniae isolates from liver abscess.