Clinical characteristics and outcome of patients with enterococcal liver abscess

Epidemiology of bacteria isolated from pyogenic liver abscesses change, and an increase in enterococci has been reported in European hospitals. The aim of this study was to investigate the clinical characteristics and outcome of enterococcal PLA. We performed a retrospective analysis of patients with microbiologically confirmed PLA at three German university centers. Indicators of enterococcal PLA were determined using binary logistic regression, and survival analysis was performed using Kaplan–Meier statistics and Cox regression analysis. Enterococci were isolated in 51/133 (38%) patients with PLA. Patients with enterococcal PLA had smaller abscess diameter (4.8 vs. 6.7 cm, p = 0.03) than patients with non-enterococcal PLA, but had more frequent polymicrobial culture results. In univariate logistic regression analysis, alcohol abuse (OR 3.94, 95% CI 1.24–12.49, p = 0.02), hepatobiliary malignancies (OR 3.90, 95% CI 1.86–8.18, p < 0.001) and cirrhosis (OR 6.36, 95% CI 1.27–31.96, p = 0.02) were associated with enterococcal PLA. Patients with enterococcal PLA had a higher mortality than patients with non-enterococcal PLA (hazard ratio 2.92; 95% confidence interval 1.09–7.80; p = 0.03), which remained elevated even after excluding patients with hepatobiliary malignancies, cirrhosis, and transplant recipients in a sensitivity analysis. The increased mortality was associated with non-fecal enterococci but not in patients with Enterococcus faecalis. In this retrospective, multicenter study, enterococcal PLA was common and indicated an increased risk of mortality, underscoring the need for close clinical monitoring and appropriate treatment protocols in these patients.


Scientific Reports
| (2021) 11:22265 | https://doi.org/10.1038/s41598-021-01620-9 www.nature.com/scientificreports/ Among Gram-positive isolates, enterococci are of great clinical importance because they are naturally resistant to TGC and, in the case of Enterococcus faecium, often exhibit beta-lactam resistance. Enterococci are frequently detected in intra-abdominal infections such as cholangitis, spontaneous bacterial peritonitis, or infected pancreatic necrosis [12][13][14][15] . A recent study from a German tertiary center 9 reported that enterococci were the most common pathogen in PLA (29%), with one in three enterococci identified being resistant to vancomycin. However, the impact of enterococcal infections on mortality is controversial 16,17 , and enterococci have also been associated with mortality in intra-abdominal infections in some studies 12 .
Therefore, the aim of this retrospective cross-sectional study was to investigate the bacterial spectrum and antibacterial resistance profile in patients with PLA from three German centers and to analyze the association of enterococcal PLA with survival.

Patients and methods
Study design. To characterize the pathogen spectrum and resistance patterns in patients with PLA, data from patients who underwent microbiological testing between 2009 and January 2020 were retrospectively collected at three German university centers (Jena University Hospital, Aachen University Hospital, and Technical University of Munich). Microbiological sampling from the PLA was performed either by percutaneous puncture guided by ultrasound or computed tomography (CT), or during surgical procedures. Microbial cultures were processed and analyzed according to standard local procedures. Patients were on antibiotics prior to microbiological samplings (n = 47, 35%) or received antibiotics directly after puncture (n = 86, 65%). Medical records, including patient records, electronic health records, imaging data, laboratory data, and nursing documentation were used to retrospectively determine clinical and laboratory data. The following variables were recorded in the medical records: Age, sex, comorbidities, medication, including antibiotics and changes in antibiotic therapy, intensive care unit (ICU) treatment, endoscopic diagnostic procedures, and interventions, length of hospital stay, and mortality. Patients without microbiological growth or without calculated antimicrobial therapy were excluded from the analysis. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the internal review boards (University Hospital Jena 3783-05/13 University of Munich 614/19 S-KH and University Hospital Aachen EK125-20) and written informed consent was waived by the ethics committees of the Jena University Hospital, the ethics committee of Aachen University Hospital and the ethics committee of Technical University Munich as only routine data was used.
Gram-negative bacteria were defined as multidrug-resistant if they were non-susceptible to at least one out of three antimicrobial categories according to the interim definition of the European Centre for Disease Prevention and Control (ECDC) and the Centers for Disease Control and Prevention (CDC) 18 . In addition, methicillinresistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococcus spp. (VRE) were defined as multidrug-resistant bacteria.
Statistical analysis. Categorical variables were expressed as absolute and relative frequencies and compared using Fisher's exact test. Continuous variables were expressed as medians with interquartiles and compared using the Mann-Whitney U test as variables were not normally distributed. Indicators of enterococcal liver abscess were determined by univariate and multivariable binary logistic regression analysis. Survival analysis was performed using Kaplan-Meier statistics and Cox regression analysis. Survival data were right-censored at loss to follow-up or at 365 days, whichever occurred earlier. Multivariable Cox proportional hazards models were used to evaluate the effects of covariates on short-term survival. A significance level of 0.05 was selected for all statistical tests. Analyses were performed using IBM SPSS Statistics 27 (IBM Corp., Armonk, NY, USA), and data were plotted using Prism 8 (GraphPad, La Jolla, CA, USA).
The majority of culture results were polymicrobial (72 cultures; 54%) with a median number of two bacterial species. The most frequently isolated pathogens were from the order Enterobacterales in 92 patients (69%), followed by enterococci (51 patients, 38%) and streptococcal species ( www.nature.com/scientificreports/
Within 365 days, 12 of the patients with enterococcal PLA died compared to 6 patients with non-enterococcal PLA (p = 0.025 by log-rank test) (Fig. 2A). This difference persisted in the sensitivity analysis when only patients www.nature.com/scientificreports/ without hepatopancreatic malignancy (Fig. 2B) or only patients without hepatopancreatic malignancy, cirrhosis and transplantation (Fig. 2C) were analyzed. The univariate hazard ratio for death was 2.92 (95% CI 1.09-7.80; p = 0.03) in patients with enterococcal PLA ( Table 4). The presence of E. faecalis was associated with mortality comparable to that of patients with non-enterococcal PLA (HR 0.93; 95%-CI 0.11-7.72; p = 0.95), whereas the presence of Enterococcus faecium or non-faecalis nonfaecium Enterococcus species was associated with significantly higher mortality (HR 3.63, 95%-CI 1.34-9.83, p = 0.01) (Fig. 2D, Table 4). The associations of PLA by Enterococcus spp. and E. faecium with survival remained significant after adjustment for age and sex (Table 4).

Discussion
In this retrospective analysis from three German tertiary centers, we demonstrate that enterococci can be isolated in more than one third of patients with PLA and indicate an increased risk of mortality. The high proportion of enterococci in PLA is consistent with another recent study from Central Europe, which shows Enterococcus spp. as the most common pathogen in PLA 9 , but significantly higher than in Asian cohorts 2,7,8 .
At present, the most common route of PLA is ascending cholangitis and invasion of liver parenchyma on the ground of obstruction by gallstone disease, strictures or malignancy 19 . Cohorts from our and other European University centers 9 are often enriched for patients with underlying hepatobiliary strictures who underwent prior biliary tract interventions. In patients with malignant biliary obstruction, antibiotic exposure predisposes to enterococcal bactibilia 20 , presumably predisposing to cholangiogenic liver abscess by enterococci. In our study, 48% of patients had a bile duct stent previously or currently implanted, which can often be colonized with enterococci in addition to Gram-negative bacteria or fungi 14,21,22 . However, in contrast to our previous analysis of acute ascending cholangitis 14,21 , neither the presence of biliary stents nor previous antibiotic exposure was www.nature.com/scientificreports/ significantly associated with enterococcal PLA in our analysis. In the present study, only alcohol use disorders, cirrhosis, and hepatobiliary cancer were linked to isolation of Enterococcus spp. from PLA in univariate analysis. Considering that gut flora reach the biliary system either often by ascent from the intestine, fecal enterococcal overgrowth as seen after TGC use 23 or changes in the microbiota composition may contribute. The gut microbiota of patients with alcohol use disorders and alcoholic liver disease is often enriched in Enterococcus spp, especially when other risk factors such as proton pump inhibitors are present 24,25 . Enterococci are part of the physiological flora of the gastro-intestinal tract. It remains controversial whether enterococci have pathologic significance, as they are usually described as having low virulence 17 . In our analysis, isolation of enterococci was associated with higher mortality, even after exclusion of patients with concomitant hepatobiliary disease, suggesting a possible influence of enterococci on mortality. In contrast to other intraabdominal infections in vulnerable patients 12 , we did not observe a correlation between coverage of enterococci by empirical therapy and outcome. This could be due to the fact that PLA requires several weeks of antibiotic therapy and the duration of empiric therapy is rather short in relation to definitive therapy.
Over the last years, vancomycin-resistant Enterococcus spp. increased, accounting for 19% of all Enterococcus isolates in blood stream infections 26 and up to 33% of enterococcal isolates from PLA 9 . In our analysis, patients with PLA due to E. faecalis, which is usually susceptible to beta-lactams and vancomycin 27 , had a prognosis comparable to that of non-enterococcal PLA, whereas more resistant enterococcal species such as E. faecium with and without vancomycin resistance were associated with higher mortality. The main cause of the increasing proportion of VRE is antibiotic exposure. Therefore, from an antibiotic stewardship perspective, our results may justify deferring enterococcal-specific therapy in non-critically ill patients with PLA until culture results are available. However, this needs to be clarified in a larger, prospective, randomized cohort.
This study has some limitations. First, it is a retrospective analysis of complex and often lengthy multimodal therapeutic approaches. In particular, a structured follow-up including data on the resolution of PLA were missing in some cases. Although the phenotypic characteristics of enterococcal PLA are quite robust because only patients with microbiologically confirmed PLA were included, there is a risk of bias in the mortality analysis because approximately 50% of patients were lost to follow-up within 90 days. Second, because of the nature of a noninterventional observational study, we could not distinguish whether enterococci were causative for the increased mortality or merely a surrogate marker for sicker patients. Third, all study sites were university hospitals, so patients with biliary structures, repeated antibiotic exposure, and in an immunocompromised state were  www.nature.com/scientificreports/ likely overrepresented. However, the association between enterococci and increased mortality remained stable even after exclusion of patients with concomitant hepatobiliary disease in a sensitivity analysis. Despite these limitations, our data show that patients with alcohol consumption and malignant biliary strictures are at risk for enterococcal PLA. Enterococcal PLA, particularly PLA by E. faecium and other non-faecalis enterococci, indicates a higher risk of mortality, underscoring the need for close clinical monitoring and individualized treatment protocols in these patients.