The role of Enterococcus spp. and multidrug-resistant bacteria causing pyogenic liver abscesses

All adult patients admitted to the University Hospital Frankfurt, Germany, between January 2009 and December 2015 with the discharge diagnosis of PLA were eligible for inclusion. For identification of possible patients, the patient chart database of the University Hospital Frankfurt was systematically searched for code K75.0 or K83.0 of the International Classification of Diseases, Tenth Revision, German Modification. Cases were included if (1) one or more discrete hepatic abscess cavities were confirmed by at least one imaging modality – ultrasound (US), computed tomography (CT) or magnetic resonance imaging (MRI) – as well as (2) by either positive culture results retrieved from the abscess or resolution of symptoms after antibiotic therapy. Patients were excluded if they were younger than 18 years old, if parasitic/amoebic abscesses were diagnosed or if available data were incomplete. The local ethics committee approved this study.

Charts were systematically reviewed and information obtained was gathered in a data collection form. Information recorded included sex, age, date of admission/discharge, underlying medical condition, initial symptoms and the intake of immunosuppressant agents, antibiotics, and proton-pump inhibitors. Additionally, initial laboratory values were documented. Laboratory results were considered to be the first values obtained upon hospital admission due to PLA or within 24 h upon presentation of PLA when PLA was not the initial cause of hospitalization. Imaging reports (CT, MRI, US) were analyzed and number and size of PLAs were documented.

For conventional microbiological culture procedures, aerobic and anaerobic conditions including the use of thioglycolate enrichment medium were applied. Species identification of recovered microorganisms was performed by matrix-assisted laser desorption/ionization time-of-flight (MALDI–TOF) mass spectrometry (VITEK MS, bioMérieux, Nürtingen, Germany) and VITEK2 (bioMérieux, Nürtingen, Germany). Antibiotic susceptibility testing (AST) was done by VITEK2 (bioMérieux, Nürtingen, Germany) according to Clinical and Laboratory Standards Institute (CLSI) guidelines and/or antibiotic gradient tests (Etest), where necessary. All laboratory tests were performed under strict quality-controlled criteria (laboratory accreditation according to ISO 15189:2007 standards; certificate number D–ML–13102–01–00, valid through January 25th, 2021).

In the vast majority of cases, positive microbiological results of abscess cavity cultures were obtained. A bacterial isolate was considered to be an MDRO if it belonged to either category VRE, MRSA or MRGN. MRGN status was defined according to the German KRINKO guideline [20]. When microbiological results were available the initial empiric antibiotic treatment was assessed and considered to be adequate if the retrieved isolates were testes to be susceptible. Additionally, mycotic coinfections were documented.

The responsible physician defined the assumed cause of PLA. The therapeutic modality was classified as either surgery, percutaneous drainage (either CT- or US-guided), biliary drainage by endoscopic retrograde cholangiography (ERCP) or percutaneous transhepatic cholangiography and drainage (PTCD), or solitary medically managed. Complications and outcome, e.g. treatment on an intensive care unit (ICU), recurrence of abscesses, mortality and duration of hospitalization, were also recorded.

Finally, local hospital surveillance data was analyzed to compare rates of Enterococcus spp. and VRE causing PLA with the overall documented infection rates of Enterococcus spp. and VRE in our gastroenterology/hepatology wards between 2010 and 2015. In brief, microbiological data were extracted from the hygiene software HyBASE 6.1 (epiNET, Germany) and cross-checked by the laboratory Information system Swisslab 7.1.3 (Roche Diagnostics IT solutions, Germany). For the calculation of the relative VRE rate, E. faecium and/or E. faecium-VRE positive cultures of all microbiological specimens submitted during 2010–2015 to the laboratory were counted. Patients with an infection where both VRE and Enterococcus were isolated were counted as patients with a VRE infection only to avoid duplicates.

For statistical analysis BiAS, Version 11.03, was applied.

Group differences were calculated using the nonparametric Mann-Whitney U test (continuous variables) or Fischer’s exact (categorical variables), as appropriate. After bivariate/univariate analysis, multivariate analysis was performed by using backward selection and a P value ≥0.10 for removal from the model. Only patients with complete data for the remaining covariates were included in multivariate analysis. Sex and age were forced into the model. Odds ratios (OR) and respective 95% confidence intervals (CI) were calculated for each variable. All statistical tests were two-sided and P values <0.05 were considered to be significant.

From 130 identified charts, 86 patients (55 men and 31 women, median age 62 years [IQR 51–72 years]) matching the described criteria were included in this study (Fig. 1). Detailed clinical characteristics and laboratory results as well as symptoms of patients, underlying diseases, direct cause and abscess’ characteristics are displayed in Tables 1 and 2, respectively.

As expected from a hospital with a major hepatobiliary surgery and liver transplant center, most of the PLA were of secondary nature originating in bile duct stenosis/obstruction (31.4%), anastomosis leakage and biliary infection (both 15.1%) as well as a superinfected liver metastasis (9.3%). In 14 patients (16.3%) the cause of liver abscess remained cryptogenic.

The most frequent comorbidities included malignancies in 34.9%, diabetes in 24.4%, the presence of liver cirrhosis in 16.3%, and prior liver transplantation in 16.3% of cases. Of note, 74.4% and 17.4% of patients received proton-pump inhibitors and immunosuppressive therapy prior to PLA formation, respectively.

Microbiological cultures (blood and/or abscess cavity cultures) were set up in all 86 patients and were positive in 77 out of 86 (89.5%). Mycotic coinfections were documented in 21 cases (24.4%), mostly caused by Candida albicans (76.2% of all mycotic coinfections). The number of recovered bacterial/mycotic species per patient is represented in Fig. 2. Overall, 135 bacterial isolates were identified; a detailed overview is displayed in Table 3. In short, both gram-negative (48.9%) and gram-positive (46.7%) aerobic bacteria were frequently cultured, whereas anaerobic bacteria were identified relatively rarely (4.4%). Of note, the most common isolated bacterial species detected were Enterococcus spp. (28.9%, in total: E. faecium 26; E. faecalis 13). Patients’ characteristics with cultures positive for Enterococcus spp. and bivariate analysis are displayed in Table 4. In multivariate analysis, there was a trend to Enterococcus infections in patients taking proton-pump inhibitors (P = 0.057, OR 3.73, 95% CI 0.96–14.41). ERCP/PTCD in the last three months (P = 0.010, OR 4.23, 95% CI 1.41–12.77) and history of prior abdominal surgery (P = 0.036, OR 3.36, 95% CI 1.08–10.44) were independently associated with Enterococcus spp. infection in patients with PLA.

Summaries of antimicrobial susceptibility and resistance profiles of enterococci, gram-positive and gram-negative aerobic bacteria are displayed in Table 5. Detailed resistance profiles of individual bacterial isolates are shown in Additional file 1: Table S1, susceptibility and resistance profiles of gram-positive aerobic bacteria excluding enterococci in Additional file 2: Table S3 and those of anaerobic bacteria in Additional file 3: Table S4. Overall, 25 MDROs were identified. Most interestingly, 35.9% of all Enterococcus spp. were classified as VRE (E. faecium 46.2%, E faecalis 7.7%) including cases of teicoplanin, daptomycin-, and linezolid- (intermediate) resistance (Table 5, Additional file 1: Table S1). Importantly, although the PLA patient populations were comparable over the study period, the relative VRE rate (VRE among all Enterococci) in patients with PLA increased over time (Additional file 4: Figure S1A and B), while in other patients without PLA it did not change significantly (Additional file 5: Figure S2A and B). In addition, 12 cases of MRGN were identified (16,7% of gram-negative bacteria). Of note, approximately 10% and 20% of tested gram-negative aerobe bacteria were resistant to carbapenems and to tigecycline, respectively. Resistance rates to fluoroquinolones, broad-spectrum penicillin/β-lactamase inhibitor combinations (BSP/βLI) and cephalosporins were relatively high (20–57%, Table 5). Results of fungal isolates analysis are displayed in Additional file 6: Table S2. Next, we analyzed both patient populations with and without MDRO causing PLA. Results of bivariate analyses are depicted in Table 6. Of note, in multivariate analysis prior known MDRO (P = 0.0002, OR 30.51, 95% CI 5.15–180.78) and the use of glycopeptide antibiotics prior to culture collection (P = 0.030, OR 6.46, 95% CI 1.20–34.87) were independently associated with MDRO causing PLA in our patients.

As initial empiric antibiotic therapy, 48.8% of included patients received a carbapenem-based regimen, 22.4% of patients BSP/βLI, and 17.6% of patients a third generation cephalosporin. In 32.6% and 8.1% of patients, a glycopeptide antibiotic or tigecycline was added to initial empirical treatment, respectively. 12 patients (14.0%) were treated with an additional antimycotic agent upon diagnosis. Of note, 35.7% of patients had received an inappropriate initial empirical antibiotic treatment, as evidenced by subsequent microbiological culture results.

Only 7 patients (8.1%) were solitary medically managed. Almost all patients received a percutaneous drainage of the abscess cavity (43.0% CT-guided, 47.7% US-guided). Additional interventional ERCP/PTCD was performed in 25.6% of cases. Surgery was necessary in 9.3% of cases.

The median duration of hospitalization was 20 days. Recurrent hepatic abscesses after discharge were reported in 18 patients (20.9%).

Twenty-four patients (27.9%) required intensive care therapy. Results of uni- and multivariate analysis are depicted in Table 7. Of note, only diabetes mellitus (P = 0.048, OR 3.72, 95% CI 1.01–13.70) and mycotic coinfection (P = 0.012, OR 5.54, 95% CI 1.46–21.06) as well a carbapenem based initial empirical antibiotic therapy (P = 0.037, OR 3.73, 95% CI 1.09–12.89) independently predicted an ICU stay during hospitalization.

Fourteen patients died during their hospital stay, resulting in an overall mortality rate of 16.3%. In univariate and multivariate analysis (Table 8), increased bilirubin levels (multivariate P = 0.015, OR 0.85, 95% CI 0.75–0.97) and presence of malignancy (multivariate P = 0.041, OR 0.19, 95% CI 0.04–0.94) were significantly associated with mortality of PLA. Of note, neither the MDRO status nor the correct initial empiric therapy was related to ICU stay or survival.