Epidemiology and Resistance Patterns of Bacterial and Fungal Colonization of Biliary Plastic Stents: A Prospective Cohort Study

Background Plastic stents used for the treatment of biliary obstruction will become occluded over time due to microbial colonization and formation of biofilms. Treatment of stent-associated cholangitis is often not effective because of inappropriate use of antimicrobial agents or antimicrobial resistance. We aimed to assess the current bacterial and fungal etiology of stent-associated biofilms, with particular emphasis on antimicrobial resistance. Methods Patients with biliary strictures requiring endoscopic stent placement were prospectively enrolled. After the retrieval of stents, biofilms were disrupted by sonication, microorganisms were cultured, and isolates were identified by matrix-associated laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry and/or biochemical typing. Finally, minimum inhibitory concentrations (MICs) were determined for various antimicrobial agents. Selected stents were further analyzed by fluorescence in situ hybridization (FISH). Results Among 120 patients (62.5% males, median age 64 years) with biliary strictures (35% malignant, 65% benign), 113 double pigtail polyurethane and 100 straight polyethylene stents were analyzed after a median indwelling time of 63 days (range, 1–1274 days). The stent occlusion rate was 11.5% and 13%, respectively, being associated with a significantly increased risk of cholangitis (38.5% vs. 9.1%, P<0.001). Ninety-five different bacterial and 13 fungal species were detected; polymicrobial colonization predominated (95.8% vs. 4.2%, P<0.001). Enterococci (79.3%), Enterobacteriaceae (73.7%), and Candida spp. (55.9%) were the leading pathogens. Candida species were more frequent in patients previously receiving prolonged antibiotic therapy (63% vs. 46.7%, P = 0.023). Vancomycin-resistant enterococci accounted for 13.7%, extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae with co-resistance to ciprofloxacin accounted for 13.9%, and azole-resistant Candida spp. accounted for 32.9% of the respective isolates. Conclusions Enterococci and Candida species play an important role in the microbial colonization of biliary stents. Therefore, empirical antimicrobial treatment of stent-associated cholangitis should be guided toward enterococci, Enterobacteriaceae, streptococci, anaerobes, and Candida. To determine causative pathogens, an accurate microbiological analysis of the extracted stent(s) may be helpful.


Introduction
Obstruction of the biliary system by malignancies, anastomotic stenosis after liver transplantation, chronic pancreatitis, or gallstones that are not immediately extractable endoscopically prevents the drainage of bile fluids from the liver and gall bladder to the small intestine and results in obstructive jaundice [1][2][3]. Gold standard for palliative treatment of an obstructed bile duct is the insertion of a stent by endoscopic retrograde cholangiopancreatography (ERCP) to restore bile fluid drainage [2,3]. However, sooner or later, biliary stents become colonized by microorganisms and ultimately occluded by a sludge (Fig 1) composed of bacteria, fungi, proteins, calcium bilirubinate, calcium palmitate, cholesterol, and plant fibers, leading to recurrent cholestasis or cholangitis [1,[4][5][6][7][8][9]. The risk for stent occlusion depends on the indwelling time of the stent [1][2][3]8,10]. According to available studies, median patency of plastic stents ranges between 70 and 126 days [2,3,8,10]. Therefore, most endoscopy units perform stent exchanges at a programmed interval of three months to avoid stent occlusion [2,10,11]. Self-expanding metal stents may double the patency time as compared to straight polyethylene stents [12]. The application of antibiotics, such as ofloxacin, alone or in combination with ursodeoxycholic acid, has failed in preventing stent occlusion [13].
A broad range of studies suggest that bacterial adhesion to artificial surfaces, resulting in the formation of biofilms, drives sludge development ahead [10,[14][15][16][17][18][19][20][21][22]. A biofilm is a structured consortium of microorganisms embedded in a self-produced polymer matrix consisting of polysaccharides, proteins, and deoxyribonucleic acid (DNA). This consortium can consist of one or more species living in a sociomicrobiological way [21]. Bacterial biofilms predispose for chronic infections because indwelling microorganisms show increased tolerance to antibiotics and disinfectant chemicals as well as resist phagocytosis and other components of the body's defense system [21]. In a pioneering study published in 1988, Leung et al. perfused biliary stents with either bacterially contaminated or sterile bile fluid [14]. Scanning electron microscopy revealed a dense layer of bacteria and amorphous material on the surface of the stents which were perfused with bacterially contaminated bile. In contrast, this phenomenon was not observed in experiments using sterile bile only [14].
The clinical management of biofilm-associated infections is challenging because the microbes in the stent are not generally amenable to eradication employing antimicrobial chemotherapy [1]. Thus, the occluded device must be replaced endoscopically, which requires an elaborate and, from the perspective of the patient, stressful procedure. Although stent replacement may be repeated several times, stent lifetime tends to decrease progressively, presumably because relevant microbes that have adapted to the stent niche accumulate within the bile duct [1]. Biofilms developing on biliary stents usually consist of a broad and diverse spectrum of mixed microbial communities [1,17,19]. However, many bacteria of the gut microbiota may not presently be cultivated by standard laboratory procedures. Thus, existing data on biliary stent microbial communities may be biased and incomplete [1]. Notably, even available information on culturable bacteria is not static and shows a recent trend toward pathogens such as extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae, enterococci, and Candida species [10,[23][24][25][26][27][28]. Thus, standard antimicrobial therapy concepts for patients with cholangitis, which mainly target sensitive Enterobacteriaceae and anaerobes, need to be reevaluated [25][26][27][28].
We performed this prospective cohort study to determine the current bacterial and fungal etiology of biofilm formation in biliary stents, with particular emphasis on antibiotic therapy and thereby exerted selective pressure. A further objective was to assess the dynamics of pathogens and the epidemiology of antimicrobial resistance compared to what was found in previous studies [10,14,17,19,20,23,24].

Study population and study design
From June 2013 to March 2015, 120 patients with biliary strictures and elective or emergency stent exchanges were consecutively enrolled. Stent exchanges were performed at the interdisciplinary endoscopy unit of the Leipzig University Hospital (Leipzig, Germany) by experienced endoscopic examiners (personal record of >250 ERCPs).

Ethics approval
The study was performed in accordance with the ethical guidelines of the 1964 Declaration of Helsinki and was approved by the local ethics committee (University of Leipzig, register no. 059-13-11032013). As the study did not modify patient management, and as the data were processed anonymously, the need for informed consent was waived according to the ethics committee approval.
Interventional endoscopic procedure ERC was performed using standard videoduodenoscopes of the Fujinon ED-530 series (Fujinon, Japan). First, the position of the indwelling biliary stent(s) was documented by direct endoscopic visualization and x-ray. Afterwards, (a) stent(s) was (were) grasped by a snare or forceps and then extracted by complete retraction of the endoscope through the stomach, the esophagus, the pharynx, and the oral cavity. Subsequently, a 5F catheter was inserted into the biliary tract, and contrast fluid was injected. Depending on the results of previous and current cholangiograms, either (a) new stent(s) was (were) inserted, or the stent therapy was ended. The caliber of subsequent stent placements varied between 8F and 11.5F, as in the previous intervention.

Stent characteristics and stent preparation
All extracted stents were made of either polyethylene (straight stents from Cook Medical, Ireland, or from Boston Scientific, USA) or polyurethane (double pigtail stents from Optimed, Germany). To minimize the risk of contamination, and to provide the best possible pre-analytic conditions, 3-4 centimeters of the distal end of the extracted stent(s), previously located in the intestinal lumen, were removed using sterile scissors, and immediately thereafter, stents were immersed in a sterile nutrient solution (Brain Heart Infusion Broth; Merck, Germany). Transport to the microbiology laboratory was performed on the same day, where the stent(s) was (were) worked up according to a standardized protocol. In patients with multi-stenting, only one stent was submitted for microbiological analysis.

Definition of stent occlusion
Before starting the sonication process, an assessment of the drainage function of the stent was carried out in the microbiology laboratory. Hereby, encrusted sludge completely narrowing the stent lumen was defined as inner stent occlusion, as described previously [10].

Sonication process
In the microbiology laboratory, the prepared stent was put into a sterile container (Bandelin, Germany) and completely covered with 50 ml of sterile Ringer's solution. To disrupt the biofilm on the inner surface of the stent, the specimen was vortexed for 30 seconds and subsequently exposed to low-frequency (40 kHz) ultrasound for 15 minutes, as described previously [10,29,30]. Sonication was performed in an ultrasound bath specially designed for microbiological analysis (BactoSonic 1 ; Bandelin, Germany). Thereafter, the container was vortexed again for 30 seconds.

Microbiological analysis
Aliquots of the sonication fluid were cultivated on conventional solid media (sheep blood agar, blood agar, chocolate agar, Esculin agar, Endo agar, Columbia agar with and without gentamicin supplementation, CHROMagar™ ESBL, and Sabouraud agar) and incubated in aerobic and anaerobic atmospheres at 37°C (Heraeus Incubator Series 6000; Thermo Scientific, Germany), as described previously [10]. Aerobic cultures were incubated for at least 48 hours, and anaerobic cultures for 96 hours. All bacterial isolates were identified using a matrix-associated laser desorption/ionization time-of-flight mass spectrometer (MALDI-TOF MS; bioMérieux, France). Fungal isolates were submitted to a commercially available assimilation test (ID 32 C yeast identification system; bioMérieux, France) for species identification.

Visualization of biofilm formation
In selected cases, biofilms were visualized by fluorescence in situ hybridization (FISH) (Fig 2) in the Biofilm Center of the German Heart Institute (Berlin, Germany), as described previously [33,34].

Blood cultures
Blood cultures were collected only in patients presenting with acute cholangitis or general signs of infection.

Statistical analysis
Statistical analysis was performed using SPSS for Windows (SPSS 20.0, IBM Corporation, Armonk, New York, USA). Numerical variables were summarized as medians, and categorical variables were given as frequencies or proportions. Confidence intervals (CI) for frequencies were calculated based on binomial distribution. Categorical data were analyzed by the chisquare test or Fisher's exact test. For the comparison of two independent groups, the nonparametric Mann-Whitney U test was used. Independent predictors for the occurrence of enterococci or Candida species were identified using multiple logistic regression. For comparing the time-dependent persistence of pathogens, the log-rank test was applied. P values (two-sided) of <0.05 were considered statistically significant.
The clinical cholangitis rate (defined by the occurrence of fever, abdominal pain, jaundice, pus coming out of the papilla) was 20% (24/120; 95% CI: 13.3% to 28.3%). Three patients had multiple cholangitis episodes due to repeating stenting procedures. The risk of cholangitis was significantly higher in patients with stent occlusion (38.5% vs. 9.1%, P <0.001). Eighty-six of the 120 patients (71.7%) received antibiotic treatments during stent insertion, mainly

Detection of bacteria and fungi
All endoscopically retrieved stents showed microbial colonization, even after an indwelling time of just one day. A total of 95 different bacterial and 13 different fungal species could be detected ( Table 2). Up to seven different species were detected on a single stent (Fig 4). Polymicrobial colonization (95.8%; 95% CI: 92.1% to 98%) was significantly more common than monomicrobial colonization (4.2%; 95% CI: 2% to 7.9%) was (P <0.001 In patients with repeating stenting procedures, a clear tendency toward the establishment of the same microbial flora on consecutive stents could be demonstrated, with a concordance rate of 84% for Enterococcus faecalis, 83% for Candida albicans, 72% for Escherichia coli, and 60% for Enteroccocus faecium as the most common pathogens. The microbiological species analysis revealed no significant difference between the microbial colonization of either polyethylene or polyurethane stents, regardless of their diameters, except for Escherichia coli (P = 0.030). Further statistically significant differences were found for colonization with Candida species (67.1% vs. 49.6%, P = 0.015), viridans group streptococci   Table).

Time dependency of stent colonization
The time-dependent graphical representation of stent colonizers grouped by enterococci, Enterobacteriaceae, Candida species, and streptococci showed a significantly different distribution of enterococci versus Candida spp. (P <0.001), and enterococci versus streptococci (P = 0.005), respectively (log-rank test) (Fig 5). Stent colonization with respect to antibiotic therapy Stent colonization differed significantly with respect to the proportion of Candida species (63% vs. 46.7%, P = 0.023) in patients receiving prolonged antibiotic therapy (Fig 6). In the other groups, no significant differences were found.

Independent predictors for colonization with enterococci and Candida spp.
Logistic regression analysis (using the following independent variables: age, gender, malignant biliary strictures, preceding liver transplantation, presence of diabetes mellitus, repeated ERC, prolonged antibiotic therapy >48 hours, and length of hospital stay >7 days) revealed that repeated ERC intervention was an independent predictor for the occurrence of enterococci (odds ratio [OR] 9.75; 95% CI: 1.24 to 76. 43), and presence of a malignant biliary stricture was an independent predictor for the occurrence of Candida spp. (OR 2.55; 95% CI: 1.08 to 6.05).

Discussion
In this study, 213 stents from 120 patients with the detection of 95 different bacteria and 13 different fungal species were included in the final analysis. Compared to previous analyses with a similar design and methodology [17][18][19][20]24], our study is distinguished by the large sample size; the use of sonication facilitating a better release of pathogens from the biofilm on the inner surface of the stents [10,29,30]; the application of MALDI-TOF mass spectrometry for precise identification of microorganisms; and a detailed antimicrobial susceptibility analysis involving 28 antibiotics and seven antifungal agents. Our results reveal the highest detection rate for both enterococci (79.3% of all stents) and Candida species (55.9% of all stents) of all studies investigating patients with biliary stents [10,17,19,20,[23][24][25][26][27][35][36][37]. The occurrence of both pathogen groups was significantly associated with an increased indwelling time of stents (Fig 5). Moreover, previous antibiotic therapy seems to exert significant selective pressure as also reported elsewhere [25,27]. Thus, the stent colonization differed significantly with respect to the proportion of Candida species (46.7% vs. 63%, P = 0.023) in patients receiving earlier prolonged antibiotic therapy. Unquestionably, this pathogen shift has clinical relevance, as conventional doctrines regarding the microbiology of the biliary tract describe a predominance of Enterobacteriaceae and anaerobes, followed by enterococci and streptococci [38,39].
Preceding studies have shown a clear association between biliary stent insertion and changes in the spectrum of pathogens in the bile toward enterococci [10,23,36], as well as the increased incidence of bacterial and fungal colonization associated with pre-operative biliary drainage [37]. In a study performed by Weber et al. in Germany, enterococci were by far the most prevalent genera, reaching detection rates of >70% in patients with stent-associated cholangitis [23]. Thus, the high proportion of enterococcal species of 79.3% in our study is in line with previous reports, revealing that samples collected from patients with stents had a significantly higher incidence of Enterococcus spp. compared to bile samples from patients without stents, corroborating a "stent tropism" of enterococci [10,23,36]. Furthermore, the number of enterococci isolated from blood cultures appeared to be higher in stent-associated cholangitis episodes [23]. In this context, repeated ERC intervention was an independent predictor for the occurrence of enterococcal species (OR 9.75) in our study. The well-described risk factors of Candida in bile include biliary stenting, malignant strictures, and repeated interventions [23,26]. In such patients, prolonged antibiotic treatment apparently facilitates biliary colonization by Candida spp. [26,27], as also shown in our study. From a clinical point of view, the detection of significantly more Candida spp. (as well as viridans group streptococci) in patients with malignant disease may also reflect a certain degree of tumor-associated immunological dysfunction.
The majority of stent extractions were performed electively, and therefore, blood cultures were collected only in patients presenting with acute cholangitis or general signs of infection. Because the matching rate of blood cultures and stent isolates was 56.7% in our study, we assume that corresponding shifts in the biliary pathogen spectrum toward enterococci and Candida spp. may be clinically relevant in the case of stent-associated acute cholangitis. However, Schneider et al. already pointed out that the hypothesis that microorganisms isolated from the biofilms on biliary stent surfaces are similar to microorganisms in the bile fluid still has to be proven [10], as concomitant bile cultures were not systematically collected in most available studies and systematic microbiological typing was omitted. Negm et al. concluded in a recent study of endoscopically obtained bile aspirates that all microorganisms found in positive blood cultures were also found in bile samples [26], indicating that biliary culture results alone are as effective as positive blood cultures. In the literature, the clinical relevance of endoscopically performed bile collection for microbiological analysis is discussed controversially [10,26,[35][36][37]. In a study performed by Park et al. in Korea, in 258 bacteremic cholangitis episodes, complete agreement with blood cultures was observed only in 31% of the bile samples [35]. Fifty percent of the bile specimens showed partial agreement with the blood culture findings, and 19% of the bile cultures revealed completely different microorganisms compared to the blood cultures analyzed. The degree of coincidence between the bile and blood cultures for Escherichia coli, Klebsiella spp., enterococci, and streptococci was 71%, 53%, 35%, and 27%, respectively [35]. In our setting, the risk of cholangitis was significantly higher in patients with stent occlusion (38.5% vs. 9.1%, P <0.001). In the study performed by Schneider et al., biliary stents occluded after a median indwelling time of 70 days [10]. However, stent occlusion resulted in cholangitis or cholestasis in only 35% of the cases. Thus, stent occlusion is not always associated with clinical symptoms.
The question of the optimal stent surface with regard to the prevention of occlusion has not yet been answered satisfactorily, whereby an advantage for Teflon™ (polytetrafluorethylene) compared to other plastics could be demonstrated [6]. Several studies have shown that antimicrobial coatings on medical devices are effective against the formation of microbial biofilms [10]. Therefore, the use of special coatings, such as diamond-like carbons (DLC) or silvernanoparticles, seems to be a promising approach for reducing biofilm formation on medical implants [40,41]. On the other hand, there has been no evidence so far that modified surfaces offer effective protection against the apparently growing threat of colonization by enterococci and Candida species.
Regarding antimicrobial resistance profiles, in our environment, resistance rates to vancomycin in enterococci (0.9% in E. faecalis, and 28.6% in E. faecium), the rate of co-resistance to ciprofloxacin in ESBL-producing Enterobacteriaceae (13.9%), and the resistance rate to carbapenems in Enterobacteriaceae (0.4%) were still within feasible ranges. However, taking into account the azole resistance rate of 32.9%, significant failure rates of fluconazole-based empirical antifungal therapies for Candida infections have to be considered. It is well known that numerous interventional ERC procedures accompanied by antibiotic therapy predispose for increased antimicrobial resistance in patients with acute cholangitis [25,27], and enterococcal species are often involved in this process [42,43]. Therefore, adequate microbiological diagnosis is essential and should include analysis of the extracted stents.
Multi-drug resistance is generally on the rise, particularly among Enterobacteriaceae [44][45][46][47]. Regarding the detection of carbapenem-resistant Enterobacteriaceae (CRE) in a single patient in our study, it is important to keep in mind that biliary colonization with CRE may have enormous clinical relevance in certain patient groups, such as liver transplant recipients [48]. Taking into consideration existing hygiene recommendations for the clinical management of affected patients in Germany, we propose active surveillance and stringent contact isolation precautions in hospitals, particularly in the endoscopy unit [49]. Invasive procedures as well as the use of antibiotics should be limited to clear medical indications.
In the overall assessment, our study has limited representativeness due to its monocentric design, and it displays several limitations: Firstly, the contamination of samples during endoscopic retrieval of stents cannot be excluded. Secondly, the choice and indwelling times of polyethylene or polyurethane stents inserted in the bile duct were arbitrary. Therefore, the timing of stent extraction and the indications for stent placement were not standardized. Thirdly, clinical selection bias cannot be excluded because it was a non-randomized study, and the results of the microbiological analysis are certainly influenced by the use of antimicrobial agents.

Conclusions
We conclude that enterococci and Candida species play an important role in the microbial colonization of biliary stents. Therefore, empirical antimicrobial treatment of stent-associated cholangitis should be guided toward enterococci, Enterobacteriaceae, streptococci, anaerobes, and Candida. To determine causative pathogens, an accurate microbiological analysis of the extracted stent(s) may be helpful. Assessing the presence of pathobionts versus symbionts in the biliary tract is an important task of future clinical studies.
Supporting Information S1 Dataset. The study data base is available for download. (XLSX) S1 Table. Microorganisms isolated from biliary stents in relation to disease and stent characteristics.
(DOC) analysis tools: AM JM ACR. Wrote the paper: CL. Critical revision of the manuscript: CL KW JF AM NL TB JM AH ACR.