Clinical Features and Prognosis of Gas-forming and Non-gas-forming Pyogenic Liver Abscess: A Comparative Study

Background Gas-forming pyogenic liver abscess (GFPLA) accounts for up to 30% of all pyogenic liver abscess (PLA) cases. However, little is known of the differences in clinical features and prognosis between GFPLA and non-GFPLA. Aim This retrospective study compared the clinical features and prognosis of GFPLA and non-GFPLA. Data of 392 patients with PLA treated from 1 January 2007 to 31 December 2016 were reviewed. GFPLA was defined as gas in the abscess. Liver abscesses were considered non-GFPLA (n = 326) or GFPLA (n = 66). The clinical features and outcomes of patients with GFPLA were compared relative to that of patients without GFPLA. The groups were similar in gender ratio, age, drinking, and coexistences. is the most in pus or blood culture of patients with either GFPLA or non-GFPLA, while the abscess in GFPLA is larger. The GFPLA group had a higher incidence of sepsis and longer hospital stay compared with the non-GFPLA. Both GFPLA and non-GFPLA can be cured by timely and effective antibiotics alone or antibiotics combined with percutaneous drainage or surgical drainage. However, early and adequate abscess percutaneous drainage may be the optimal treatment for GFPLA.

GFPLA is associated with past abdominal surgery, especially hepatobiliary surgery.
Patients with a history of abdominal surgery should be monitored more closely in the early stage of the PLA. GFPLA has high rates of sepsis and long hospitalization. It needs to be recognized as a distinct clinical entity.

Background
Pyogenic liver abscess (PLA) is a common infectious disease of the liver, accounting for 80% of all liver abscesses [1]. Compared with the United States and Europe, the rate of incidence of PLA is higher in Asia, specifically about 17.6/100,000 in Taiwan and 1.1-3.6/100,000 in mainland China [1,2]. The incidence of PLA has increased steadily in recent years, as it is associated with diabetes mellitus, an aging population, increasingly aggressive surgical management of hepatic, biliary, and pancreatic disorders, and the wide use of immunosuppressive agents in patients with transplant and cancer [3]. Gasforming PLA, that is, with gas in the abscess based on ultrasound or computed tomography (CT) imaging data ( Figure), may constitute as much as 30% of PLA cases [4].
GFPLA was first reported by Smith in 1944 [5]. Patients with GFPLA have a high risk of mortality, and for this it has gained attention [4] .The most common pathogen associated with PLA is Klebsiella pneumoniae, a Gram-negative rod-shaped bacterium with a high rate of fatality [6]. In patients with diabetes, for example, the high glucose levels may provide gas-forming microorganisms with a favorable environment for forming gas through mixed acid fermentation. In an acidic environment (pH < 6), mixed acid is metabolized by formic hydrogenlyase, resulting in gas formation [7]. Moreover, the microangiopathy associated with diabetes markedly impairs transport of catabolic end products (i.e., gas) away from the lesion, and thereby results in gas accumulation [8,9].
However, few reports of PLA have focused on the differences in clinical features and prognoses of patients with or without gas-forming abscesses (GFPLA and non-GFPLA, respectively). This retrospective study compared the clinical features and prognoses of patients with GFPLA or non-GFPLA in our hospital.

Methods
This study was approved by the Ethics Committee of First Affiliated Hospital of Xi'an Jiaotong University (No: XJTU1AF2015LSL-057). The need for patients' informed written consent was waived due to the retrospective nature of the study.

Study population
In all, 438 patients with PLA were admitted to First Affiliated Hospital of Xi'an Jiaotong University from 1 January 2007 to 31 December 2016. Candidates with the following were excluded: 39 with incomplete history, 1 with amoeba liver abscess, 3 with tuberculosis liver abscess, and 3 younger than 18 years. The remaining 392 patients were apportioned to either the GFPLA or non-GFPLA group, according to the presence (n = 66) or absence (n = 326) of gas in the abscess, respectively. Gas-forming PLA, that is, with gas in the abscess based on Ultrasound or/and computed tomography (CT) image findlings. All PLA patients received second or third generation cephalosporins combined with metronidazole or antibiotics selected according to the drug sensitivity of pus culture. Depending on the number, size, liquefaction, and separation of the abscess, response to antibiotics and the professional opinion of the doctor, the patients were given antibiotics alone, antibiotics combined with percutaneous drainage or surgical drainage [10,11].

Data collection
A retrospective comparison and analysis was conducted of the clinical features and prognosis of patients in the GFPLA and non-GFPLA groups. Clinical data included demographic data (age and gender), lifestyle habits (smoking and drinking), coexistences, medical histories, symptoms and signs at admission, and changes during hospitalization.
Laboratory data included blood routine examination and liver, renal, coagulation functions. The number, size and site of the abscess were determined by ultrasound, CT, or both. Bacterial culture included blood culture within 24 hours after admission and pus culture was executive through puncture-guided by ultrasound or CT. Treatment included antibiotics alone, antibiotics combined with percutaneous drainage or surgical drainage.
Complications included sepsis, septic shock, acute kidney injury (AKI), acute respiratory distress syndrome (ARDS), spontaneous rupture of abscess, systemic inflammatory response syndrome (SIRS), pleural effusions, and portal venous thrombosis. The clinical effect was based on the days of temperature normalization after admission, length of hospital stay, and mortality in hospital.

Statistical analysis
Continuous variables were expressed as mean ± standard deviation, and categorical variables as absolute numbers and percent frequencies. Differences between continuous data were analyzed using Student's t-test. Differences between categorical data were analyzed with the chi-squared test or Fisher's exact test, as appropriate. All statistical analyses were performed with SPSS version 22.0 (IBM, Armonk, NY). A 2-tailed P value < 0.05 was considered statistically significant.

Demographic data, lifestyle habits, coexistences and medical histories
Among the overall study population of 392 patients, there were 66 with GFPLA and 326 with non-GFPLA, aged 57.0 ± 13.7 and 56.7 ± 13.3 years, respectively ( Table 1). The corresponding ratios of men-to-women in the two groups were 38:28 and 185:141. The GFPLA and non-GFPLA groups were statistically similar regard to gender ratio, age, smoking, drinking, and coexistences (hypertension, diabetes, hepatobiliary malignancy, lithiasis, liver cirrhosis, and viral hepatitis). A history of abdominal surgery was significantly more common in patients with GFPLA compared with non-GFPLA (62.1% vs. 41.7%; Table 1; P = 0.002), especially hepatobiliary surgery (50% vs. 30.3%; Table 1; P = 0.024). A greater percentage of the non-GFPLA group had no history of surgery compared with the GFPLA (58.3% vs. 37.9%; Table 1; P = 0.002). There was no significant difference in the medical history of PLA between the two groups (Table 1; P = 0.086). Table 2 showed that there were no significant differences in the symptoms and signs of fever, chill, abdominal pain, nausea, vomit, fatigue, body temperature, respiratory rate, or mean arterial pressure (MAP) between the two groups (Table 2). However, the heart rate of the GFPLA group (88.3 ± 13.1 bpm) was significantly faster than that of the non-GFPLA group (84.5 ± 12.8 bpm; Table 2; P = 0.028).
Coagulation function: As showed in table 2, there were no significant differences in activated partial thromboplastin time (APTT) and fibrinogen (FIB) between the two groups.

Image findings
As showed in table 2, the two groups were similar with regard to percentages of single or multiple abscesses. Specifically, in the GFPLA and non-GFPLA groups there were, respectively, 52 (78.8%) and 239 (73.3%) cases of single abscess, and 14 (21.2%) and 87 (26.7%) cases of multiple abscesses. However, the maximum diameter was significantly higher in the GFPLA (7.8 ± 2.8 cm) compared with the non-GFPLA (6.5 ± 2.8 cm; Table 2; P = 0.001). There were 57 patients in the GFPLA group and 288 in the non-GFPLA group.
Among them, in the GFPLA and non-GFPLA groups there were, respectively, 11 (19.3%) and 35 (12.2%) located in the left lobe of the liver, and 39 (68.4%) and 216 (75.0%) were located in the right lobe. Seven (12.3%) GFPLA and 37 (12.8%) non-GFPLA were found in both sides of the liver lobe. And abscess site was similar in both groups.

Microbiological characteristics
Pus culture: There were similar in pus microbiological proportions (Table 3; P > 0.05), number of pus bacterial species (Table 3; multiple bacteria; 6.6% vs. 16.7%; P = 0.054) and pus negative rate (Table 3; 34.4% vs. 29.2%; P = 0.491) between two groups. Most of the positive bacteria in pus culture were K. pneumoniae, followed by Escherichia coli. Blood culture: There were similar in blood microbiological proportions (Table 3; P > 0.05), number of blood bacterial species (Table 3; multiple bacteria; 3.1% vs. 6.3%, P = 0.335), most of the positive bacteria in pus culture were K. pneumoniae, followed by Escherichia coli, which were consistantly with results of pus culture between two groups.
However, the Non-GFPLA group had a higher blood negative rate than GFPLA group (Table   3; 77.1% vs. 56.2%, P = 0.017).  (Table 4). In the GFPLA group, these corresponding treatments were administered to 12 (18.2%), 45 (68.2%), and 9 (13.6%) patients. A significantly higher percentage of patients in the GFPLA group received antibiotics combined with percutaneous drainage compared with the non-GFPLA, but a significantly lower percentage in the GFPLA underwent either conservative treatment or surgical drainage (Table 4; P = 0.016).

Outcomes:
The groups were similar in days for temperature normalization after admission (Table 4; 7.2 ± 6.3 vs.8.1 ± 6.0 d; P = 0.298). However, the patients of the GFPLA group experienced a significantly longer hospital stay compared with the non-GFPLA (Table 4; 19.6 ± 11.6 d vs. 16.2 ± 8.8 d; P = 0.029). There were no deaths in either group during hospitalization.

Discussion
The clinical features and prognosis of GFPLA has not been sufficiently elucidated. Few studies have investigated the role of gas-formation in PLA [4,9,12,13]. In the present study, we retrospectively compared the clinical features and outcomes of GFPLA and non-GFPLA. The analysis found that GFPLA was associated with previousabdominal surgery, especially hepatobiliary surgery. GFPLA was associated with a high rates of sepsis and long hospitalization. However, the clinical features, the prevalence of diabetes, and microbiological characteristics were similar to that of the non-GFPLA patients. More importantly, there were no major differences in the overall outcomes between the GFPLA and non-GFPLA patients.
The analysis showed that GFPLA does not differ from non-GFPLA by gender ratio, age, rates of smoking, drinking, hypertension, diabetes, hepatobiliary malignant diseases, lithiasis, liver cirrhosis, and viral hepatitis. However, GFPLA was more prevalent in patients with a history of abdominal surgery, especially hepatobiliary surgery. Lee et al. [9] reported that patients with past gastrectomy may be prone to GFPLA due to alteration of gut flora. Although our current data cannot confirm that abdominal surgery is an independent risk factor for GFPLA, it does suggest an association.
The symptoms and signs of the GFPLA and non-GFPLA groups were no statistically comparable, except that the heart rate of the patients with GFPLA was significantly higher. It has been reported that an elevated heart rate is associated with increased mortality and cardiovascular risk in patients with no significant heart disease [14]. This suggests that the higher heart rate in GFPLA may portend a poorer prognosis.
A link between type 2 diabetes and GFPLA remains controversial. In some studies, the presence of diabetes and poorly controlled blood glucose was the single strongest risk factor for GFPLA [12,[15][16][17], but Yang et al. [13] reported no significant difference in blood glucose levels between these patients with or without GFPLA. The latter is consistent with the present study, which found no significant difference in the prevalence of diabetes between the two groups. However, the GFPLA group had significantly lower levels of lymphocytes, hemoglobins, platelets, and albumin compared with the non-GFPLA.
Imaging analysis showed that there was no significant difference in the site and number of abscesses between the GFPLA and non-GFPLA groups. Yet, the abscesses of the GFPLA group were usually larger (i.e., by maximum diameter). This finding is consistent with Chou et al. [12], and may be due to the expansion of the abscess with infection and gas formation.
In the present study, the composition and relative predominance of bacterial species in the blood and pus cultures were similar. K. pneumoniae was the most common causative organism, followed by Escherichia coli. This is consistent with recent literature [18]. This result may be related to the aging social structure of the population and the increased incidence of diabetes and tumors. Several studies have shown that the hyperglycemic environment and abnormal intima of diabetic patients is beneficial to K. pneumoniae colonization, and facilitates the spread of K. pneumoniae through the blood to cause PLA [19]. The negative rate of blood culture in the non-GFPLA group was higher than that of the GFPLA. The reason for this is not known.
The rate of pleural effusions in the GFPLA group was significantly higher than that of the non-GFPLA. This may be due to the adjacent anatomical relationship between the chest and the liver. The severe infection of the liver directly stimulates the diaphragm, to aggravate pleural effusions. It may also be caused by the partial destruction of liver function, the increase of portal vein pressure, and the obstruction of lymphoid reflux in the liver abscess.
In the past, abscess drainage was mainly accomplished through surgery. However, percutaneous aspiration or drainage has now become the standard of practice [20,21]. In the current study population, 68.2% of the patients with GFPLA and 48.8% of the patients with non-GFPLA underwent antibiotics combined with percutaneous drainage, which is consistent with the above findings. However, the rate of antibiotics combined with percutaneous drainage was significantly higher in the GFPLA, and the rates of antibiotics combined with surgical drainage and antibiotics alone were lower. This may be because the patients with GFPLA showed a higher positive rate of bacteria in blood culture. It was thus easier to select sensitive antibiotics for curing the disease.
In this study, no death occurred in any patient during hospitalization, and there was no significant difference in overall short-term prognosis between the two groups. Earlier studies reported that patients with GFPLA have a high mortality rate [13,22]. However, due to differences in the etiology, diagnosis, treatment, and prognosis of PLA, the incidence and mortality of GFPLA still needs investigation. Moreover, whether the presence of gas in the abscess has a decisive effect on the prognosis of PLA needs to be confirmed.
Several limitations of this study need to be considered. First, the data was from only a single center, and most of the patients reside in Shaanxi Province of China. Considerable differences in etiology, treatment, and outcomes of PLA have been revealed in recent studies from different parts of the world [23-25]. Therefore, our findings need to be verified by multicenter research. Secondly, this is a retrospective study and subject to selection bias and residual confounding. Finally, we only investigated the short-term results of PLA in this study. This is because the underlying disease would significantly influence the long-term outcomes of the patient. To evaluate GFPLA for higher incidence and mortality, a prospective study with propensity score matching is necessary. GFPLA preferentially occurs in patients with a history of abdominal surgery, especially hepatobiliary surgery. K. pneumoniae is the most common pathogenic bacteria in pus or blood culture of patients with either GFPLA or non-GFPLA, while the abscess in GFPLA is larger. The GFPLA group had a higher incidence of sepsis and longer hospital stay compared with the non-GFPLA. Both GFPLA and non-GFPLA can be cured by timely and effective antibiotics alone or antibiotics combined with percutaneous drainage or surgical drainage. However, early and adequate abscess percutaneous drainage may be the optimal treatment for GFPLA.

Consent for publication: Not applicable.
Availability of data and material: All data generated or analysed during this study are included in this published article. * Reported as n (%), unless shown otherwise.    Figure 1 Abdominal CT findings of GFPLA. Arrows: GFPLA.