Clinical characteristics and pathogens of pediatric pyogenic liver abscess in Beijing Children’s Hospital, China

Background: Data on pyogenic liver abscess (PLA) of children in China have been limited. Method: We retrospectively reviewed PLA cases from January 2017 to June 2019 at Beijing Children’s Hospital. Clinical features and pathogens were analyzed. Results: We diagnosed 15 PLA patients in our center. Median age of onset was 3.2 years and the median diagnostic time was 9 days. Four patients (26.7%) had underlying diseases and seven patients (46.7%) had multiple organ involvement. The sensitivity achieved 100% when ultrasound and contrast-enhanced CT scan were combined use to diagnose PLA. Two patients were defined pathogen through metagenomic next-generation sequencing (mNGS). Finally, eleven patients (73.3%) were cured during follow-up time and four patients (26.7%) dead. Compared to the survival group and the death group, the patients in death group were more likely to have multiple organ involved (100 vs. 27%, P < 0.05) and bilateral lobe occupied ( P < 0.05) . Conclusions: PLA often occurs in patients with underlying disease. Multiple organs involved may lead to poor prognosis. mNGS can be a new method to detect pathogen.

worldwide. Similarly, upward tendency was shown in China [8] . Patients with underlying diseases are tending to get PLA. Diabetic mellitus (DM), cirrhosis, immune-compromised state are the well-known predisposing factors [10] .
But so far, most studies focused on adults but seldom children, which resulted in the limited data in our country. Here, we present a single-center study of PLA and analysis clinical characteristics and pathogens in pediatric population.

Methods
In this retrospective study, we reviewed data from children (younger than 18 years old) who admitted to Beijing Children's hospital (National Center for children's Health, China; A 970-bed tertiary pediatric hospital) and diagnosed PLA from January 2017 to June 2019. All of the patients met with at least one of the following inclusion criteria [11] : (1) Presence of abscess in the liver on imaging examinations; (2) Confirmed by aspiration for bacteriology studies. We collected the information of the risk factors, clinical presentations, laboratory tests including pathogen results, radiological findings, management and outcomes of PLA patients.

Statistical analysis
Mean and standard deviations (SDs) are shown when distributions were confirmed normal; median and interquartile ranges (IQRs) are reported otherwise. The categorical variables were compared using the Chi-square test or Fischer's exact test, as appropriate. Continuous variables within two groups were compared using the independent t-test for parametric data and the Mann-Whitney U test for non-parametric data. P values < 0.05 were considered statistically significant. All of the statistical analyses were conducted using SPSS 23.0 (SPSS Inc., USA)

Results
Baseline characteristics of the patients Fifteen children were diagnosed as PLA from January 2017 to June 2019, including 8 males (53.3%) and 7 females (46.7%), the median age of onset was 3.2(range 0.17, 5.6)years, the minimum was 17day-old and the median diagnostic time was 9 interquartile range 5, 22 days. Five (33.3%) patients had been treated in intensive care unit (ICU). The clinical characteristics of patients with PLA are concluded in Table 1. Predisposing factors Among 4 out of 15 patients (26.7%) who had underlying diseases, one was juvenile idiopathic arthritis with operation of abdomen and another one had acute lymphoblastic leukemia, both of whom had been received a long-time treatment of glucocorticoid and immunosuppressive drugs. And one patient was DM, another one was primary immunodeficiency.

Pathogen findings
Aspirations were done in 5 patients (33.3%) and one of them was confirmed the infection of K. pneumonia through pus culture, moreover another two were identified Streptococcus intermedius (unique reads 3) and Bartonella henselae(unique reads 92) separately by the pus of metagenomic next-generation sequencing (mNGS). Thirteen patients were taken blood culture, Staphylococcus aureus (Methicillin-resistant staphylococcus) was isolated from one patient and Pseudomonas aeruginosa (carbapenemase positive) was isolated from another patient. The rest had negative results of blood and pus culture.

Imaging findings
All patients performed ultrasound and nearly one half (46.7%) performed the contrast-enhanced computed tomography (CT) scan simultaneously. The detail features were shown in Table 1. Six patients (40.0%) had single abscess, the rest was multiple. Eleven out of 15 patients (73.3%) had unilateral involvement and 8 patients (53.3%) occupied right lobe. Thirteen patients presented with low hypoechoic lesions in ultrasound, two patients were normal. Seven patients (46.7%) who underwent contrast-enhanced CT scan showed lesions of low density with edge intensified (Fig. 2).

Therapy and outcomes
All patients received intravenous antimicrobial therapy, nine patients (60.0%) with the treatment of single agent initially (Fig. 3), three of which changed antimicrobial treatment after obtaining the result of pathogen. Among one patient received single agent of meropenem and then combined linezolid at the time detected Streptococcus intermedius by mNGS. And one patient who was proved Bartonella henselae changed from cefatriaxone to rifampicin together with azithromycin. Initial treatment of one patient was cefoperazone/sulbactam and pathogen of liver pus culture was negative, but K.pneumonia was isolated from pus culture of intra-abdominal and then changed meropenem companied with vancomycin. Finally eleven patients (73.3%) took a favorable turn and were confirmed to be cured through imaging findings during follow-up time. Four patients (26.7%) discharged without treatment and then dead. Compared to the survival group and the death group, the patients in death group were more likely to have multiple organ involved (100 vs. 27%, P < 0.05) and bilateral lobe occupied (P < 0.05).

Discussion
HA is invaded by microbes via bile ducts, blood stream (hepatic artery and portal vein) or by contiguous spread [11,12] and PLA is accused of the majority causes of HA [1] . Numerous studies show the risk factors which predispose patients to PLA, including immune-compromised state (immunodeficiency, chemotherapy and etc.), DM and so on [13][14][15] . In our research, two patients took glucocorticoid and immunosuppressive drugs and one patient had primary immunodeficiency, which were belonged to a state of immune-compromising so that it was susceptible to form abscess. Also, another patient was considered severe combined immunodeficiency due to the significant decreased in serum immunoglobulin and T cell subsets, however it was unable to confirm the diagnosis that his parents refused to do genetic test. Furthermore one patient had DM which was noted as a condition that immune system was wakened [15] , thus getting infected easily.
Almost all patients (86.7%) had fever and a fair of patients (26.7%) presented abdominal pain, which could be the most common clinical features and basis to make the diagnosis of PLA [9] . Meanwhile several patients had cough due to the stimulation of diaphragm through liver abscess nearby and according to some studies that patients complain of non-specific symptoms such as fatigue, anemia and weight loss [16] . Which is easy to misdiagnose during clinical assessment and requires concerning on PLA when treat these patients, especially for immunosuppressed populations.
At the initial state of disease, most patients(86.7%) had elevated level of CRP, eleven patients(84.6%) had increased level of PCT and twelve patients(80%) presented with leukocytosis, all of these indicated the infection and especially for bacterium, which was proved in various studies of PLA [1,16] .
More than half (53.3%) had abscess in the right lobe, could be correlated with blood flow direction and volume [1] . Two patients with small multiple abscesses were detected by CT scan, but undetected by ultrasound. The sensitivity achieved 100% when ultrasound and contrast-enhanced CT scan were combined use. Actually ultrasound was more widely used in clinical work due to universality of technology, ease of operation and higher cost performance, even as a tool to perform ultrasoundguild aspiration. It has been reported contrast-enhanced CT scan was done in PLA patients [16][17][18] , but owing to the potential risk of liver or renal function impaired and the consideration of economic burden, ultrasound became one of the main imaging examination method to diagnose PLA.
In our study, K.pneumonia was isolated from a patient with DM through pus culture. Another patient whose pus culture of liver abscess was negative, whereas K.pneumonia was isolated through pus culture of intra-abnormal abscess, which could be suspected that the pathogen of PLA in this patient was K.pneumonia probably. In the past few decades, S.aureus is the most common pathogen isolated from PLA in children all over the world [8,9] . Whereas in recent years, with the depth understanding of disease and improvement of laboratory technology, the pathogen of PLA have changed. In a metaanalysis of pathogen distribution with PLA in China [8] , Klebsiella spp was the most common pathogen and mostly occurred in DM. K.pneumonia strain is mostly K1 or K2 serotype. Neutrophil phagocytosis and chemotaxis may be impaired due to the poorly controlled of blood glucose [12,15,19] , leading to infected with K.pneumonia easily.
Culture-based techniques are the golden standard, and varies studies advanced more pathogens can be isolated through pus culture [18,20,21] . In our study the positive rate of blood or pus culture was not very high, it could be attribute to sample collection after prior antibiotic treatment of prehospitalization. Meanwhile, it is of difficulty to cultivate in restricting conditions and takes a long time to get results which are regarded as limitation of culture and isolation of pathogen.
It is worth noting that in our study patients infected with Streptococcus intermedius(unique reads 3) and Bartonella henselae(unique reads 92) were detected by mNGS, a new detection technology in recent years and presenting with promising result in the field of pathogen diagnosis [22] . A study in our center about evaluation of mNGS for the pathogenic diagnosis showed the ability to identify pathogens from abscess and less affected by prior antibiotic [23] . Applying this technology in clinical practice to gain more experiences is our direction in the future.
All patients received antibiotics on account of susceptibility and experience, the monotherapy with meropenem or third-generation of cephalosporin was the main regimen in our study (53.3%), which conformed to the pathogen distribution of PLA in our country [8] . In one patient, the liver abscess was enlarged after receiving antibiotics only and then finished aspiration. It is recommend [24] that all patients should be given antibiotics and evaluated therapy response in 48 hours, drainage is necessary to be done when no response, incomplete response or abscess enlarged. If US showed the abscess rupture or impending rupture, large with the location of left lobe, requiring intervention because of associated abdominal lesion, urgent drainage was demanded. In our study, the rate of complications was 46.7%, 57%( 4/7) of which were dead. A research indicated about 15.7% of HA patients develop complications, which can be accused of most death [19] .
This study has some limitations. It was a single center, hospital-based retrospective design, which introduces the possibility of unrecognized biases. Abscess biopsy was done in seldom patients and due to the usage of antibiotics before admitted to our hospital, the positive rate of pathogen was not very high. Further multicenter studies are necessary to enrich the generalizability of PLA in children.

Acknowledgments
We would like to thank the Medical Records and Statistics Room, Beijing Children's Hospital, which provided hospital discharge data.

Funding
This work was supported by Beijing Hospital Authority "Dengfeng" Talent Training Plan (DFL 20181201). The funders had no role in study design, data collection and analysis, or preparation of the manuscript.

Availability of data and materials
The datasets during and/or analyzed during the current study are available from the corresponding author on reasonable request.