Introduction
Pyogenic liver abscess (PLA) is a common and severe intraabdominal infection that could be caused by bacterial or fungal organisms. Because of delayed diagnosis and inappropriate treatment, before the 1970s, the mortality rate of liver abscess through nonsurgical or surgical approaches was as high as 83.00% and 28.00%, respectively [
1]. Since the 1990s, surgical drainage has become the mainstream therapeutic option. Nowadays, advanced imaging techniques and therapeutic modalities have greatly improved patients’ prognosis. Most liver abscess could be treated by non-surgical measures, and the death rate has decreased from 70% to about 16% [
2-
4].
In terms of causative pathogens,
K. pneumoniae had emerged as the most common pathogenic bacteria worldwide instead of
Escherichia coli in the past ten years [
5-
8].
K. pneumoniae, a Gram-negative facultative anaerobic bacillus, is the most important opportunistic pathogen apart from
E coli. Formate hydrogenlyase endows
K. pneumoniae with a gas generating capacity, and the fatality rate of gas-forming pyogenic liver abscess (GFPLA) is dramatically high [
9]. Besides, extensive use of broad spectrum antibiotics has led to the appearance of extended spectrum β-lactamase (ESBLs) producing strains. These
K. pneumoniae strains are resistant not only to the third generation cephalosporins, but also to aminoglycosides and fluoroquinolones thus easily ending with refractory infection [
10,
11]. The high mortality and antibiotics-resistant rates make the treatment of
Klebsiella pneumoniae liver abscess (KLA) always a thorny problem. Undoubtedly, early detection and proper therapy are essential; however, consensus has not been achieved in many detailed issues. As the gold standard of pathogenic diagnosis, bacterial culture is useless to make early judgment about KLA for its delay in time and low culture-positive rate; therefore, clinical data seem to be more critical for physicians to make timely diagnosis.
Our research aims to find evidence for early diagnosis of KLA through a comprehensive comparison between KLA and NKLA patients on the aspects of underlying diseases, clinical characteristics, laboratory data and imaging findings, and to provide rational therapeutic modalities of KLA through analysis of treatment responses to different medical interventions in KLA patients.
Materials and subjects
Enrolled patients
One hundred and ninety-seven inpatients, diagnosed as having PLA in Zhongshan Hospital of Fudan University, Shanghai, China, from March 2001 to September 2009, were enrolled in this research. Selected diagnostic criteria are as follows: (1) clinical symptoms such as fever, chill and abdominal pain; (2) radiological findings of liver abscess either by ultrasonography or computed tomography (CT); and (3) percutaneous liver aspiration as part of the therapy and abscess aspirate culture for pathogenic diagnosis. Demographic and medical information of all patients were collected. The medical information included underlying diseases, clinical characteristics (symptoms and signs), laboratory parameters, microbiological culture results, radiological findings (size, location and number of the abscess), and patients’ responses to treatment. Underlying diseases and related risk factors included diabetes, biliary diseases, liver cirrhosis, chronic hepatitis B, hepatic adipose infiltration, history of abdominal surgeries, and malignant tumors. Laboratory parameters were obtained as follows: hemoglobin, leucocyte count, percentage of neutrophils, C-reactive protein (CRP), albumin, total bilirubin (TB), alanine transaminase (ALT), alpha fetoprotein (AFP), and blood glucose. The values on admission were recorded for analysis. Responses to different percutaneous liver aspiration modalities were evaluated by comparison before and a week after treatment in terms of leucocyte count, percentage of neutrophils, abscess size, and time length to defervescence (time from receiving abscess drainage until body temperature decreased to below 37°C). All data were collected and checked carefully by two independent researchers.
Percutaneous liver aspiration
All enrolled patients had undergone ultrasonography (conventional abdominal ultrasonic probe, without puncture-guiding device, 3.5 MHz) guided percutaneous liver aspiration (18G percutaneous needle; “J” guide-wire). Before liver puncturation, a multiple-section scanning was performed to choose the best entry point and puncture angle. The PTC needle was pricked into the abscess cavity and fixed in an appropriate location, then the needle core was pulled out, and a 30-mL syringe was linked to aspirate pus. Normal saline or antibiotics (metronidazole and gentamicin) were used to wash the abscess cavities three to five times. Some patients had a catheter implanted and retained for subsequent treatment.
Microbiology and antimicrobial susceptibility testing
The specimens (blood or liver abscess aspirate) were sent for Gram staining, bacterial cultures (standard aerobic and anaerobic diagnostic methods), and tests for antimicrobial susceptibility. An automatic identification system for Gram-negative rods was used for species identification. The culture results were described as follows: patients who had at least one code for a microorganism (culture-positive) or without a code for specific microorganism (culture-negative); and those with only one code (monomicrobial) or with more than one code (polymicrobial). Antimicrobial susceptibility was determined by using the Bauer-Kirby disk-diffusion method on the Mueller-Hinton agar medium, as recommended by the guidelines of Clinical and Laboratory Standards Institute. After the cultures of blood and/or liver abscess aspirate had been obtained, initial empirical broad-spectrum antibiotic treatment was administered intravenously, and adjusted as necessary when the results of the culture and susceptibility test were known.
Imaging investigation
Ultrasonography (TOSHIBA-240, MEDISON-6000, manual operation) or CT (Siemens Sensatation 16 CT, GE LightSpeed VCT, ADW4.3 workstation) examinations were performed before the percutaneous liver aspiration. Before examination, patients were asked to drink 800-1000 mL water. The scan range was from the top of diaphragm to the lower pole of the kidney. The arterial phase scanning proceeded after the hepatic plain scan was finished, and the delay time reached 22 s (iodic contrast medium, concentration: 300 mg/mL, total volume: 90 mL, injection rate: 5 mL/s). All CT reports were made independently by two experienced radiology specialists in Zhongshan Hospital.
Intervention group
According to the liver aspirate culture results, patients who were only infected by a single pathogen (monomicrobial patients) were divided into two pathogen groups for analysis: patients with K. pneumoniae liver abscess (KLA group) and those with non-Klebsiella pneumoniae liver abscess (NKLA group). To evaluate the treatment responses of KLA patients to different liver aspiration modalities, the KLA group was further divided into four treatment subgroups: group I (percutaneous liver aspiration), the pus was aspirated after successful puncturation, but without antibiotics washing or retained catheter; group II (aspiration plus antibiotics flushing), pus was aspirated with subsequent antibiotics washing, but without postoperative catheterization; group III (aspiration plus retained catheter), pus was aspirated with postoperative catheterization, but without drug washing; group IV (aspiration plus antibiotics flushing and retained catheter), pus was aspirated with both antibiotics washing and postoperative catheterization. The patients took a postoperative bed rest for 2-3 h with application of routine hemostatic drugs. The intravenous antibiotics therapy continued for more than three days.
Statistical analysis
All collected data were analyzed using the Statistical Package for the Social Sciences (version 16.0; SPSS, Chicago, IL, USA). Univariate analysis was performed by Student’s t test and Mann-Whitney U test for continuous variables, depending on whether the data had a normal distribution; chi-squared test for categorical variables; and Fisher’s exact test when more than 20% of the expected frequencies were less than 5. Data are expressed as mean±standard deviation. The result was considered to be significant if P<0.05.
Results
Microbiological characteristics of enrolled PLA patients
Of 197 liver abscess patients, 20 cases (10.16%) were culture-negative, and 177 (89.84%) were culture-positive, including monomicrobial (n=162: KLA, 106; NKLA, 56) and polymicrobial (n=15). Among culture-positive patients, the most common organisms isolated were K. pneumoniae (60.10%), E. coli (13.99%), enterococci (7.25%), and Bacillus aeruginosus (6.74%). Other bacteria were Staphylococcus aureus, Citro bacter freumdii, Enterobacter cloacae, etc. (Table 1).
Demographic data and underlying diseases
The male-to-female ratio was 2.7∶1.0 in KLA group and 2.3∶1.0 in NKLA group, while the mean age was 58.02±2.40 (range, 13 to 76; median 57) years in KLA group and 56.04±3.52 (range, 24 to 82; median 55) years in NKLA group. There were no significant differences in age and gender between the two groups (P>0.05). Compared to NKLA, more KLA patients were associated with diabetes mellitus (53.77% vs 25.00%, P=0.001), hepatic adipose infiltration (16.04% vs 5.36%, P=0.029); however, fewer KLA patients were related to abdominal surgeries (16.98% vs 35.71%, P=0.001), chemo-radiotherapies (1.89% vs 10.71%, P=0.021), or malignant tumors (2.83% vs 12.50%, P=0.033) (Table 2). There were no statistical differences in the morbidities of biliary diseases, liver cirrhosis, or chronic hepatitis B between the two groups.
Clinical presentation and laboratory parameters
The most common symptoms in KLA and NKLA patients were fever and/or chill with no significant difference between the two groups (73.58% in KLA vs 67.86% in NKLA, P=0.442). Compared to NKLA, the common symptoms such as abdominal pain (57.14% in NKLA vs 40.57% in KLA, P=0.044), fatigue (46.43% in NKLA vs 19.81% in KLA, P=0.001), and hepatomegaly (14.29% in NKLA vs 4.72% in KLA, P=0.033) were much milder in KLA patients. There were no significant differences in other symptoms/signs such as vomiting, anorexia, highest body temperature, or hepatic percussion pain/tenderness between the two groups (Table 3).
On admission, leucocyte count, percentage of neutrophils, alanine transaminase (ALT) and total bilirubin (TB) and CRP levels of PLA patients were frequently elevated, while the albumin and hemoglobin levels were frequently decreased (Table 3); however, no significant differences were found between KLA and NKLA groups. In consistence with the fact that KLA was more likely to occur in patients with diabetes, patients in the KLA group usually had a higher blood glucose level on admission than NKLA (7.84±0.36 mmol/L vs 5.76±0.30 mmol/L, P=0.001).
Imaging findings
Ultrasonographic findings from 162 monomicrobial PLA patients indicated that patients either with KLA or NKLA had a higher frequency of solitary abscess (82.19% in KLA vs 81.08% in NKLA, P=0.535) in hepatic right lobe (65.09% in KLA group vs 69.64% in NKLA group, P=0.422). Besides, KLA had a higher incidence to display its gas-forming feature than NKLA (25.47% vs 10.71%, P<0.001) by ultrasonography. And the two groups showed no significant difference in abscess size (diameter, 73.85±2.40 mm in KLA vs 73.77±3.22 mm in NKLA, P=0.341) or compartment occurrence in abscess (38.67% in KLA vs 35.71% in NKLA, P=0.711) (Table 4).
Of the 162 monomicrobial PLA patients, 110 had undertaken abdominal CT examination. By conventional plain scan, KLA patients were more likely to have abscess with blurring edge (71.23% vs 40.54%, P=0.002) and gas cavity (32.88% vs 13.51%, P=0.039) in comparison with NKLA; however, the two groups shared the same frequency of pleural effusion (34.25% vs 29.73%, P=0.633) and biliary pneumatosis (9.59% vs 8.11%, P=1.000). During the artery enhancement period of CT examination, a higher incidence of septal enhancement was much more frequent in KLA than in NKLA (41.10% vs 16.22%, P=0.009). However, no significant difference was found in the presentation of combined pneumobilia (9.59% vs 8.11%, P=1.000), pleural effusion (34.25% vs 29.73%, P=0.633), or portal venous phase delayed enhancement between the two groups (Table 4).
Treatment and clinical outcomes
All of the 106 enrolled monomicrobial KLA patients received parenteral antibiotics during the early period of fever. The commonly used antibiotics scheme to treat KLA was metronidazole/ornidazole-based anti-anaerobic agents in combination with fluoroquinolones and/or third-generation cephalosporins. Hydrocarbon mold vinyl and aminoglycoside antibiotics could also be used to treat KLA. Two patients (1.89%) complicated with fungal infection were cured by fluconazole. Besides, intravenous nutrition was extensively used during KLA treatment, of which, 37 cases (34.9%) were given extra liver-protecting therapy.
According to different liver aspiration regimens, the KLA patients (n=106) were further divided into four treatment subgroups: group I (percutaneous liver aspiration, n=24); group II (aspiration plus antibiotics flushing, n=52); group III (aspiration plus retained catheter, n=24); and group IV (aspiration plus antibiotics flushing and retained catheter, n=6). Here, metronidazole and gentamicin were used to wash abscess cavities after liver aspiration. KLA patients who had a relatively large abscess with a diameter above 30 mm were treated with catheter drainage rather than simple percutaneous liver aspiration. One case in group III died because of poor general condition and was excluded from later analysis. The other KLA patients were improved obviously after treatment and discharged with stable body temperature.
To assess the therapeutic effect of antibiotics flushing on KLA, the treatment outcomes of group I (n=24) and group II (n=52) were analyzed by using laboratory and ultrasonography data before and one week after abscess drainage. However, no significant difference was found in leucocyte count decrease, decline in percentage of neutrophils, reduction of abscess diameter and time length to defervescence between groups I and II. To assess the therapeutic effect of retained catheter on KLA, the outcomes of group III (n=23) and group I (n=24) /group II (n=52) were compared. It turned out that the retained catheter had its advantage in reducing the diameter of the abscess (Table 5).
Follow up
One hundred and five KLA patients (one dead case was excluded) continued oral antibiotics therapy after discharge and were advised to take reexamination at months 0.5, 1, 3 and 6 to adjust antibiotics regimens. The median follow-up time of patients after discharge was four months (range: one to nine months). Recovery of liver abscess was defined as normal laboratory parameters and body temperature with disappeared abscess by ultrasonography, while recurrence of liver abscess was defined as the development of new clinical and radiological changes after discharge. Ninety of the 105 KLA patients had received follow up at out-patient clinics, and the rate of loss to follow up was 14.28%. Two patients with diabetes in group I had experienced recurrence of liver abscess: one at the fourth month was infected by K. pneumoniae again, and the other at the eighth month infected by E. coli. The average healing time (days from abscess drainage to recovery) of other KLA patients without recurrence was 40.12±5.67 days, and no statistical difference was found between the four treatment groups.
Discussion
PLA is usually caused by bacteria or opportunistic pathogens from hepatic artery, portal vein, and intestinal tracts. The normal liver can kill small quantities of invasive bacteria and suppress their reproduction because of its abundant blood supply and strong defense system. Immune dysfunction or bacterial overload will dramatically increase the incidence of liver abscess.
E. coli was recognized as the most common pathogenic bacteria in liver abscess by biliary or portal infection, followed by
Anaerobius streptococcus. While in cryptogenic abscess or those disseminated by hepatic artery, staphylococcus was the most common, especially
S. aureus. However, nowadays,
K. pneumoniae has gradually become one of the most important pathogens in liver abscess. Our research indicated the culture-positive rate of
K. pneumoniae in liver abscess was 60.10%, which coincides with recent data “60%-70%” from other researches [
6,
7,
10].
K. pneumoniae, as one of the most important opportunistic pathogens, was not common before the 1970s, and usually invaded patients in a state of immune suppression and caused severe infection [
12]. Nowadays, the extensively-used broad-spectrum antibiotics, especially third-generation cephalosporins, have contributed greatly to the prevalence of
K. pneumoniae, and
K. pneumoniae has become an important pathogen in hospital-acquired infection [
10,
11,
13]. However, most PLA patients began with a fever and had received antibiotics therapy at an early stage in community hospitals, and this would inevitably result in the deviation of culture results. During recent years, ESBLs-producing strains have been isolated in several laboratories including China [
14-
16], and the ESBLs-producing incidence of
K. pneumoniae strains was quite high: 20.00% of 1248
K. pneumoniae strains isolated from a large-scale hospital in Lebanon produced ESBLs and the highest incidence was 38.40% in its ICU department [
13]; the incidence of isolated
K. pneumoniae strains was even as high as 51.00% in the ICU of Huashan Hospital, Shanghai, China. Undoubtedly, these ESBLs-producing
K. pneumoniae strains were resistant to penicillin. Besides, they were also resistant to third-generation cephalosporin, aminoglycosides and fluoroquinolones. Therefore, early diagnosis and reasonable application of antibiotics at an early stage are critical to
K. pneumoniae infection, for it can not only improve the efficacy of treatment but also prevent fatal antibiotic resistance.
At the beginning of the 20th century, PLA was the most common complication of appendicitis [
1]. With improvement in intraoperative asepsis, biliary diseases had gradually become the major cause of NKLA. It was reported that KLA was much more related to diabetes. Also, reports about PLA as a complication of minimally invasive surgeries such as chemoembolization and stent implantation for malignant tumors have been increasing [
3,
9,
17,
18].
Diabetic patients are the high risk group of PLA, especially KLA. On one hand, diabetic patients with disorders of glucose degradation could not provide enough energy to leucocytes, thus leading to damage of their immune defense function. On the other hand, high blood glucose level provides a favorable microenvironment for bacterial reproduction. Meanwhile, diabetic patients are likely to have complications of vascular abnormalities which could cause local oxygen deficiency, and this not only favored growth of anaerobic bacteria but also impaired leucocytes’ oxygen-dependent bactericidal capacity [
19,
20]. Therefore, diabetic condition is quite suitable for facultative anaerobic bacteria such as
K. pneumoniae.
Our study confirmed that 39.60% of PLA patients were complicated with diabetes mellitus, and this incidence further manifested as 53.77% in KLA and 25.00% in NKLA (P<0.05). This indicated that diabetes-associated PLA was probably KLA. Moreover, KLA was more related to fatty liver than NKLA (P<0.05), and it could be easily explained as “high blood glucose itself is a high risk of fatty liver.” Our study did not find a peculiarity in the pathogen of biliary PLA. Also, the pathogens of PLA secondary to abdominal surgeries (appendectomy, gastrectomy, colectomy, etc.) or transcatheter hepatic arterial chemoembolization (TACE) were mainly NKLA, including E. coli, P. aeruginosa, and enterococcus. NKLA often occurred one month to two years after surgeries because these bacteria were the major pathogenic organisms that usually caused postoperative wound infections.
The PLA included in this study often had an acute onset with fever (highest temperature 42.0°C), abdominal pain, fatigue, anorexia and so on. Physical signs including hepatomegaly, liver tenderness and hepatic percussion pain were dramatic. Our study revealed that, compared to NKLA, the clinical characteristics including abdominal pain, fatigue and others were much milder in KLA (P<0.05), which was probably due to the complication of diabetic neuropathy in the above 50.00% patients of the KLA group, thus making the clinical symptoms of KLA mild and atypical.
Laboratory data from our study showed that leucocyte count (highest level: 55.0 × 109/L) and percentage of neutrophils in PLA patients dramatically increased, while hemoglobin and albumin frequently decreased, probably because PLA, as one of consumptive diseases, often happened in the elderly. Elevated ALT and TB also suggested that PLA impaired liver function. The fasting blood glucose level on admission was the only laboratory parameter with a significant difference between KLA and NKLA, further supporting the connection between K. pneumoniae infection and diabetic condition.
With an average diameter of 73.85±2.40 mm, solitary abscess in the right lobe of the liver was more frequent in PLA by ultrasonography, which implied that the pathogenic bacteria might derive from the portal system. Compared to NKLA, CT examination further revealed three important features of KLA (Figs. 1 and 2): (1) KLA had a higher incidence to develop gas-forming abscess, which is consistent with previous data [
4,
9]. 32.88% of KLA patients had gas-forming abscess, and 75.00% of these patients had diabetes complication. Therefore, diabetic condition could be considered as an important risk factor of gas-forming KLA. (2) KLA had a higher incidence to have a blurring edge, which also coincides with relevant data. This was probably related to the strong invasive ability of
K. pneumoniae, for it was reported that distant dissemination such as pulmonary abscess was prone to happen in KLA [
21]. (3) KLA had a higher incidence of having septal enhancement. Most of the patients included in our study received CT examination at the stage of abscess formation; therefore, dramatic septal enhancement reflected strong inflammatory response locally caused by
K. pneumoniae. In addition, Kim
et al. [
4] reported that “cluster sign” was quite rare in KLA for it was easier to break the wall of abscess causing the integration of multiple abscesses with its powerful invasive ability. Our study did not find this feature probably because all included KLA patients underwent CT scan at the very early stage of liver abscess, and ultrasonography, instead of CT, was used to monitor changes of abscess dynamically in our hospital.
Therapy of liver abscess should be chosen under comprehensive consideration of the patients’ general condition, stage of PLA, size and location of abscess and complications. Because of high incidence of the multi-drug resistance of
K. pneumoniae, choosing an effective therapy for KLA is particularly important. With continuous development of modern imaging techniques, ultrasonography-guided percutaneous liver aspiration or catheter drainage has occupied a dominant position in the treatment of PLA [
3,
11,
13]. Here, through a retrospective analysis on the efficacy of different intervention procedures, we found several meaningful points during KLA treatment: (1) Extra antibiotics flushing after abscess drainage was meaningless. No significant difference was found between groups I and II. Many physicians believe that local antibiotics flushing could wash residual pus and benefit the absorption of abscess, and it was reported that most patients could acquire stable body temperature one to two days after abscess drainage with antibiotics flushing. This is inconsistent with our data. The average time length to defervescence of group II (aspiration plus antibiotics flushing) was 5.35±0.84 days in our study, and one case with normal chest X-ray had pulmonary abscess which was suspected as abscess metastasis after liver aspiration plus antibiotics flushing. Although the internal reason was unclear, the suddenly-increased pressure by flushing abscess cavities would lead to the rupture of abscess and its distant metastasis, which could be one of the considerations. The antibiotics flushing showed no dominance in improving the efficacy of KLA treatment in our study, which also supports that the intravenous administration of antibiotics was sufficient to cure
K. pneumoniae infection. (2) Percutaneous catheter drainage was a better interventional procedure for KLA treatment. Compared to group I (mono-percutaneous liver aspiration) and group II (aspiration plus antibiotics flushing), catheter drainage showed its advantage in raising therapeutic response of KLA patients, especially at the reduction in abscess diameter. For the existence of separations inside the abscess in KLA, single liver aspiration could not achieve adequate drainage while the retained catheter is useful for continuous drainage of residual pus or newly-formed abscess and for subsequent aspiration if necessary. Moreover, retained catheter could reduce patients’ pain from repeated puncturations and relevant medical costs.
Conclusion
This study further confirmed that K. pneumoniae has become a major pathogen of KLA. Compared to NKLA, KLA is more likely to occur in the elderly with diabetes or fatty liver. The clinical symptoms are not typical. KLA is prone to be a solitary abscess in the right lobe of the liver with gas-forming feature, blurring edge and septal enhancement during arterial phase. Combined with sufficient application of intravenous antibiotics, timely percutanous catheter drainage is better for KLA treatment. All these findings will contribute greatly to early diagnosis and treatment of KLA. However, inevitably, our retrospective analysis had some data loss and discrepancy in physicians’ choice of antibiotics during KLA treatment, and rigorously-designed perspective studies are still needed to better guide the diagnosis and treatment of KLA.
Higher Education Press and Springer-Verlag Berlin Heidelberg
PDF
(217KB)
