A retrospective study of pyogenic liver abscess focusing on Klebsiella pneumoniae as a primary pathogen in China from 1994 to 2015

Pyogenic liver abscess (PLA) is a common intra-abdominal infection in adults. In this study, we aim to explore demographic and clinical characteristics of PLA focusing on Klebsiella pneumoniae (K. pneumoniae) induced PLA (KP-PLA) in mainland China. A retrospective review of medical records from all patients with KP-PLA admitted to a tertiary teaching hospital over a 21-year period (1994–2015) was performed. Among 296 PLA cases with confirmed culture-positive data, K. pneumoniae was revealed as the predominant pathogen (n = 189, 63.9%), followed by Escherichia coli (n = 39, 13.2%). Strikingly, KP-PLA patients had a higher incidence of metabolic disorders, such as diabetes mellitus (49.7% vs. 36.4%, P = 0.027; odds ratio (OR): 1.725; 95% confidence interval (CI): 1.061–2.805), hypertension (38.1% vs. 19.6%, P = 0.001; OR: 2.520; 95% CI: 1.439–4.413), and fatty liver (32.3% vs. 14.0%, P = 0.001; OR: 2.923; 95% CI: 1.564–5.462) than those with non-K. pneumoniae induced PLA (non-KP-PLA). Moreover, patients with KP-PLA had higher susceptibility to septic metastatic infection at distant sites compared to those with non-KP-PLA (10.6% vs. 3.7%, p = 0.038). Our results indicate that K. pneumoniae is the predominant pathogen of PLA in mainland China. KP-PLA is frequently diagnosed in patients with metabolic diseases and has a higher risk for septic metastatic infection.


Variables
PLA (n = 296) N (%) or mean ± SD Comparison of patients with KP-PLA and non-KP-PLA. Given that K. pneumoniae was found to be the leading causative pathogen in PLA, we further explored the clinical characteristics of KP-PLA (Table 3). Comparison of patients with KP-PLA and non-KP-PLA revealed no significant differences in age or gender. KP-PLA was found to be preferentially located in the right hepatic lobe (71.7% vs. 57.9%, p < 0.05) and predominately cryptogenic in origin (51.9% vs. 17.8%, p < 0.001) compared with non-KP-PLA, which commonly developed secondary to underlying hepatobiliary or colorectal diseases, malignancy and intra-abdominal trauma or surgery.  Although it seemed more patients with KP-PLA were subjected to blood-culture (81.0% vs. 70.1%, p = 0.033), we observed a lower positive rate as compared with non-KP-PLA (33.3% vs. 56.0%, p < 0.001). On the contrary, KP-PLA showed a higher positive rate in liver abscess aspirate culture (97.6% vs. 90.1%, p = 0.020). As for treatment strategies, more KP-PLA patients underwent radiologic-guided percutaneous drainage comparing to patients with non-KP-PLA (92.8% vs. 84.4%, p = 0.035). Strikingly, medical chart review of KP-PLA patients revealed significant higher incidence of underlying metabolic disorders, such as diabetes mellitus (49.7% vs. 36 16.8%). Patients with KP-PLA also tended to spend a longer time in ICU (9.0 ± 12.5 days vs. 4.0 ± 2.7 days) and hospital (18.7 ± 13.1 days vs. 18.5 ± 12.5 days). But above findings did not reach statistical significance.

Comparison of patients with KP-PLA and E. coli induced PLA (EC-PLA).
There were 39 cases of EC-PLA, which accounted for 36.3% of non-KP-PLA. Patients with either KP-PLA or EC-PLA had almost the same age and a similar gender ratio. Unlike KP-PLA, of which almost half were primary and cryptogenic in origin, all EC-PLA cases were secondary to intra-abdominal trauma or surgery, and hepatobiliary diseases. As listed in Table 6, EC-PLA patients more often had a history of hepatobiliary surgery (76.9% vs. 25.4%, p < 0.001) and underlying hepatobiliary diseases (97.4% vs. 46.6%, p < 0.001) as compared with KP-PLA, implying an association between EC-PLA and concomitant hepatobiliary pathology abnormalities. Moreover, we also observed that more KP-PLA patients had underlying metabolic diseases, such as hypertension (38.1% vs. 17.9%, p = 0.016) and fatty liver (32.3% vs. 10.3%, p = 0.006) compared with EC-PLA. Overall, the prognosis appeared to be better in patients with KP-PLA than patients with EC-PLA as far as in-hospital mortality (5.9% vs. 12.8% is concerned, but this did not reach statistical significance. Primary KP-PLA and invasive liver abscess syndrome caused by K. pneumoniae. To further explore the clinical features of KP-PLA, we analyzed the subgroup primary KP-PLA and highlighted the emergence of invasive liver abscess syndrome associated with K. pneumoniae. As shown in Table 4, among primary KP-PLA reviewed in present study, 14 cases (14.3%) developed metastatic infection at distant sites and considered as invasive liver abscess syndrome caused by K. pneumoniae. Of those patients, 2 cases (14.3%) developed meningitis, 1 case (7.1%) had brain abscess, 3 cases (21.4%) grew lung abscess, 1 case (7.1%) had kidney abscess, 4 cases (28.6%) developed septic shock, 6 cases (42.9%) required ICU support, and 3 cases (21.4%) died eventually. We also noticed that 4 (4.1%) out of 98 cases of primary KP-PLA had a history of PLA, so these 4 cases were considered as recurrence of primary KP-PLA, and the range of interval of recurrence was from 7 to 96 months.

Discussion
To the best of our knowledge, this single center retrospective study on PLA, in particular KP-PLA, covered the longest time span of all the studies reported thus far (May 1994 to Dec 2015), and provided new insights into the demographics and clinical features of patients with KP-PLA in mainland China. At our institution, 1 out of 2400 and 3700 hospital admissions were due to PLA and KP-PLA, respectively. We observed that the crude annual incidence of PLA and KP-PLA increased significantly from 1994-2004 to 2004-2015, which could be due to larger number of inpatients and improved early diagnostic techniques. In short, PLA and KP-PLA are common health problems in mainland China, which consistent with the recent epidemiological trends observed in other countries and regions 6,12,13,[19][20][21][22][23][24][25] .
K. pneumoniae was considered as the main pathogen of PLA in last three decades. Most clinical studies on KP-PLA were from the Asia-Pacific region such as Taiwan, Hong Kong, Singapore and Korea [6][7][8][9]11,12,25,26 . To date, few study cohorts were reported from mainland China. This retrospective review of all cases of PLA at our hospital confirmed K. pneumoniae as the predominant pathogen of PLA followed by E. coli. Among KP-PLA cases, approximately half were primary PLA and cryptogenic in origin, implying that K. pneumoniae can invade a previously healthy liver. Whilst a report from Taiwan only discovered 1 from 160 cases of KP-PLA that was secondary to intrahepatic duct stones 11 , in our cohort a substantial number of KP-PLA cases were reported in patients with a history of intra-abdominal surgery or underlying hepatobiliary diseases. Therefore, there are considerable region differences with respect to causes that underlie development of KP-PLA. Variations in demographic composition of the study population may also contribute to different findings. Non-KP-PLA mainly occurred as secondary to intra-abdominal surgery or hepatobiliary diseases involving E. coli, Enterococcus, Staphylococcus, Streptococcus, Pseudomonas aeruginosa infection etc., consistent with previous reports 2 .
Next, we explored the patient characteristics that were associated with PLA and KP-PLA. Our data supported the views that KP-PLA mainly occurred in middle-aged men, mostly located in right hepatic lobe and had the Scientific RepoRts | 6:38587 | DOI: 10.1038/srep38587 tendency to develop distant metastasis infection 13 . More importantly, our study demonstrated that KP-PLA was significantly associated with metabolic disorders including hypertension, diabetes mellitus and fatty liver. We also showed that metastatic infection was more frequently in KP-PLA patients with hypertension, diabetes mellitus    11,26,28 . Up to date, limited studies observed the potential association between KP-PLA and hypertension or fatty liver, but the hidden reason still required further clarification 6,15,29 . Apart from that, quite a few PLA patients had a history of gastrointestinal cancer (41/296, 13.9%), implying a correlation between PLA and gastrointestinal cancer. Consistent with our data, Sung et al. 30 showed that gastrointestinal cancer had a 4-fold higher incidence among PLA patients as compared to controls. Hence, further evaluation to rule out potential malignancies in PLA patients should be recommended in clinical practice. Extrahepatic metastatic infection is a devastating complication for KP-PLA patients. Metastatic KP-PLA patients were in relatively severe conditions and more often admitted to ICU, but even so, about 20% patients still died eventually. To eliminate the effect of variation in demographic composition of KP-PLA study population, we traced distant metastatic infection both in cases of KP-PLA and primary KP-PLA. In this study, the incidences of extrahepatic metastatic infection in KP-PLA and primary KP-PLA were 10.6% and 14.3%, respectively, consistent with reports from other countries and regions (8-28%) 11,13,14,31 . A small number of primary KP-PLA cases (n = 2) developed metastatic meningitis, a rare but life-threatening complication, whose rate of death in present study was as high as 100%. Septic endophthalmitis, a major complication reported in studies from Taiwan and other regions 14,26,27,31,32 , was not observed in current study. Such discrepancy may be caused by geographic differences and inadequate follow-up of patients. In total, the in-hospital mortality in KP-PLA and primary KP-PLA with metastatic infection were 20% (4/20) and 21.4% (3/14). Given the potentially catastrophic metastatic complications associated with KP-PLA, more attention should be paid to its detection and management among clinicians.
Given the prevalence of PLA and its severe complications, particularly KP-PLA, there is a need for early detection and appropriate treatment strategy for this disease. In our study, pus culture was found to be more sensitive in detecting K. pneumoniae than blood culture. Thus, it is preferable to obtain pus samples from either fine needle  aspiration or abscess drainage for bacteria identification prior to the use of empirical antibiotics. For those that needed abscess drainage, radiology-guided percutaneous drainage was the preferred method. As an alternative, surgical drainage was often performed in PLA patients secondary to hapatobiliary diseases or as salvage therapy in case percutaneous drainage failed. The advances in less invasive interventional radiology and endoscopic techniques 33 will further improve the treatment of PLA. In this study, almost all of the K. pneumoniae isolates from patients with KP-PLA were susceptible to all the antibiotics tested except for ampicillin despite the large amount of antibiotics consumption in mainland China, which may lead to the change of gut microbiota. Considering the poor prognosis of metastatic infection complication of KP-PLA, it is of great importance for clinicians to detect the extrahepatic infection earlier and choose appropriate antibiotics with higher tissue concentration.
In present study, we observed that the blood culture positive rate (33.3%) in KP-PLA was lower than previously reported 46-61.1% by other studies 12,20,31 . We speculated that the collection time, the frequency of sampling, the amount of blood drawn for culture and number of bottles within 24 hours all might contribute to the relatively low positive rate. Furthermore, empirical antibiotics usage prior to the collection of blood culture might be another factor contributing to the low culture positive rate. We have retrospectively reviewed these patients' medical charts and found some patients have visited local clinics or hospitals before being referred to our hospital. Thus, we cannot exclude the possibility that some of these patients were already on antibiotics prior to blood collection. This study has several limitations. First, this study is a single center, retrospective analysis that might cause selection bias in terms of patient population admitted to this hospital and recall bias related to medical history. Second, strain specificities were not determined due to limited awareness of K. pneumoniae in mainland China, whilst virulence factors such as aerobactin rmpA genes, and capsular type K1 or K2 antigen have been shown to contribute to metastatic infection and severe disease 7,8,13,26,27 . Third, high percentage of monomicrobial liver abscess in Non-KP-PLA may be attributed to the presence of predominant bacteria such as E. coli, low positive   rate of anaerobes, and may reflect inappropriate culture techniques for microorganism identification. Finally, quite a number of patients involved in present study developed septic shock and required ICU support, some predictor tools such as APACHE II score, Glasgow Coma Scale (GCS) and Sequential Organ Failure Assessment (SOFA) can be applied to further explore the severity and prognosis of this infectious disease. In summary, PLA is a common infectious disease that requires hospitalization in mainland China. K. pneumoniae is the leading pathogen of PLA, and KP-PLA patients are at higher risk of developing metastatic complications and in-hospital mortality. Metabolic disorders, including hypertension, diabetes mellitus and fatty liver are common underlying conditions in patients with KP-PLA. Invasive liver abscess syndrome caused by K. pneumoniae is an emerging clinical syndrome with distant metastasis infection in mainland China and clinicians should be highly alert of its clinical characteristics to optimize patient management.

Study population.
We performed a single-center retrospective study by compiling the inpatient medical records of all consecutive cases of PLA at the Sir Run Run Shaw Hospital, School of Medicine, Zhejiang University, Hangzhou, China, from May 1994 to December 2015. This hospital is a 2400-bed tertiary teaching hospital, which was opened in May 1994 and has an approximate annual adult admission of 46,000 during 21 years. It was the first public hospital in mainland China accredited by the Joint Commission International, a US-based, World Health Organization-authorized organization for medical quality evaluation. The list of patients with liver abscess was generated by accessing the ICD-9 (International Classification of Diseases, 9 th Revision) code (572.0) of the hospital discharge database from May 1994 to April 2013 and ICD-10 (International Classification of Diseases, 10 th Revision) code (K75.0) from May 2013 to December 2015. The initial date of discharge with a diagnosis of liver abscess was defined as the index date.
Data collection. Patient data were retrieved from paper medical records (from May 1994 to May 2010) and electronic medical charts (from June 2010 to December 2015). The patients with PLA included in our study met the following criteria: 1) older than 18 years; 2) PLA was the major cause of the hospitalization. 3) had identifiable bacteria on blood or pus culture; We excluded patients who were 1) hospitalized due to concomitant diseases, 2) developed PLA only as a complication, 3) diagnosed with amoebic liver abscess, infected liver cyst or hepatobiliary cancer diagnosed by pathological results after discharge, and 4) previously diagnosed with PLA 30 days prior to their admission date to include only patients with newly-onset PLA.

Ethics Statement. The institutional review board of Sir Run Run Shaw Hospital, School of Medicine,
Zhejiang University approved this retrospective study protocol and waived from the need for written informed consents from these patients. Patient characteristics. The patients' demographics, clinical presentations, laboratory values, radiographic findings, microbiological characteristics, treatment strategies and outcomes were collected and analyzed. LOHS, admission to ICU, LOIS (if applicable), and invasive procedures performed were documented and summarized.
Definitions. PLA was defined as the presence of liver abscess detected by imaging technologies, together with typical clinical manifestations of infection, such as fever and right upper abdominal pain. Primary PLA was defined as PLA that occurs in the absence of a history of intra-abdominal trauma or surgery, including medical interventions such as transarterial chemoembolization (TACE) and radiofrequency ablation for treatment of hepatocellular carcinoma (HCC), underlying hepatobiliary diseases such as cholecystitis, gallstones, cholangiolithiasis, and malignant obstruction of bile ducts, underlying colorectal diseases such as polyps, adenoma and cancer, and immunosuppression due to cancer chemotherapy. On the contrary, secondary PLA was defined as PLA with above comorbidities. According to the identified pathogenic bacteria isolated from blood or liver abscess aspirate, PLA was separated into KP-PLA and non-KP-PLA, with the latter was further classified as EC-PLA and those that was neither KP-PLA nor EC-PLA. Septic metastatic infection was defined as extrahepatic infectious complications including Central Nervous System (CNS), eye, lung, muscular and skeletal system and urinary system infection, such as meningitis, brain abscess, endophthalmitis, lung abscess, necrotising fasciitis, kidney abscess and other type infection. Invasive liver abscess syndrome was defined as primary KP-PLA with septic metastatic infection at distant sites 13 .
The definition of septic shock was updated by the Sepsis Definitions Task Force in 2016 and patients with septic shock were identified with a clinical construct of sepsis with persisting hypotension requiring vasopressors to maintain MAP (mean arterial pressure) ≥ 65 mmHg and having a serum lactate level > 2 mmol/L despite adequate fluid resuscitation 34 . Diabetes mellitus was defined as a 2 h-plasma glucose ≥ 200 mg/dl, a fasting plasma glucose ≥ 126 mg/dl or a random plasma glucose ≥ 200 mg/dl combined with classic symptoms of hyperglycemia 35 . Hypertension was defined as a systolic blood pressure ≥ 140 mmHg or a diastolic blood pressure ≥ 90 mmHg 36 . Fatty liver was defined as the presence of hepatic steatosis based on an ultrasound or CT scan 37 . In-hospital mortality was defined as death from any cause during hospitalization. Size of abscess was defined as the maximal cavity diameter of abscess or largest abscess when there were multiple abscesses.
Microbiology laboratory procedures. All microbiology samples including blood and pus were processed in a central laboratory for culture. The Dade Microscan Walkaway (Dade Behring, US) (from May 1994 to Dec 2002) and the VITEK 2 system (bioMe ́ rieux, Marcy l'Etoile, France) (from Jan 2003 to Dec 2015) were used to identify the bacterial isolates and perform susceptibility testing. Anti-microbial agents tested included ampicillin, ampicillin-sulbactam, amikacin, aztreonam, ciprofloxacin, cefuroxime, cefepime, cefotaxime, ceftazidime, piperacillin, piperacillin-tazobactam, cefoperazone-sulbactam, tetracycline, sulphamethoxazole-trimethoprim (SMZ-TMP) and imipenem. The extended-spectrum beta-lactamases (ESBLs) phenotype for all the collected isolates was confirmed by both a double-disk synergy test (DDST) and phenotypic confirmatory disc diffusion test (PCDDT) according to the manual of Clinical and Laboratory Standards Institute (CLSI).