Bacterial bile duct colonization in perihilar cholangiocarcinoma and its clinical significance

Abdominal infections including cholangitis represent a major problem in patients with perihilar cholangiocarcinoma (pCCA). Thus, we investigated bacterial colonization of the bile ducts and determined its impact on postoperative outcome focusing on abdominal infections. A cohort of 95 pCCA patients who underwent surgery between 2010 and 2019 with available intraoperative microbial bile cultures were analyzed regarding bile duct colonization and postoperative abdominal infection by group comparisons and logistic regressions. 84.2% (80/95) showed bacterial colonization of the bile ducts and 54.7% (52/95) developed postoperative abdominal infections. Enterococcus faecalis (38.8%, 31/80), Enterococcus faecium (32.5%, 26/80), Enterobacter cloacae (16.3%, 13/80) and Escherichia coli (11.3%, 9/80) were the most common bacteria colonizing the bile ducts and Enterococcus faecium (71.2%, 37/52), Enterococcus faecalis (30.8%, 16/52), Enterobacter cloacae (25.0%, 13/52) and Escherichia coli (19.2%, 10/52) the most common causes of postoperative abdominal infection. Further, reduced susceptibility to perioperative antibiotic prophylaxis (OR = 10.10, p = .007) was identified as independent predictor of postoperative abdominal infection. Bacterial colonization is common in pCCA patients and reduced susceptibility of the bacteria to the intraoperative antibiotic prophylaxis is an independent predictor of postoperative abdominal infections. Adapting antibiotic prophylaxis might therefore have the potential to improve surgical outcome pCCA patients.

Perihilar cholangiocarcinoma (pCCA) is the second most common primary liver tumor and associated with a dismal overall oncological prognosis and high perioperative morbidity and mortality 1,2 . Extensive liver resection with radical lymphadenectomy and vascular resection is the current gold standard for the treatment of patients with resectable disease showing encouraging survival rates up to 60% after 5 years in various selected cohorts [3][4][5][6] . In contrast, palliative treatment modalities such as systemic chemotherapy result in a significantly inferior oncological outcome 7 . However, due to its direct proximity to major vascular structures of the liver hilus, surgical therapy of pCCA patients remains challenging and often displays significant perioperative mortality rates exceeding 10% 6,[8][9][10] .
Given the extraordinarily high perioperative morbidity and mortality in these individuals, preoperative endoscopic biliary drainage (EBD) or percutaneous biliary drainage (PBD) are commonly used to release the pressure in the bile duct system, improve the liver function and promote postoperative functional recovery 11,12 . Due to preexisting conditions such as bile duct dilatation and reduced bile flow as well as the increased risk of bacterial contamination during the above-mentioned preoperative procedures, bacterial cholangitis develops in a significant proportion of pCCA patients and has been identified as an independent preoperative predictor of increased perioperative mortality [13][14][15] . Despite its potentially major impact on the perioperative outcome, the incidence and pattern of bacterial bile duct colonization have not been investigated so far. Bacterial bile duct colonization and cholangitis might also lead to postoperative abdominal infections with the risk to cause bacterial sepsis with often fulminant or even fatal complications and a reduced functional recovery of the remnant liver 16 . Previous reports have identified abdominal infections as a major source of adverse clinical outcomes, playing a key role in the development of postoperative complications, post-hepatectomy liver failure, and vascular complications 6,[8][9][10] .
Thus, we here aimed to investigate bacterial bile duct colonization in a large European cohort of pCCA patients and determine its impact on the postoperative outcome focusing on abdominal infections.

Material and methods
Patients. Between 2010 and 2019, all surgically treated patients with localized pCCA at the University Hospital RWTH Aachen (UH-RWTH) with available information on bile duct colonization from intraoperative sampling were included in this study. Informed consent was obtained from all study participants. The study was conducted in accordance with the requirements of the Institutional Review Board of the RWTH-Aachen University (EK 114/20), the current version of the Declaration of Helsinki, and the good clinical practice guidelines (ICH-GCP). All clinical data were prospectively collected and entered in an institutional database.
Staging and surgical technique. All pCCA patients who were referred to our institution for surgical treatment underwent a detailed clinical work-up as previously described 9,10 . In brief, the decision for surgery as primary treatment and the specific surgical procedure was made by an experienced hepatobiliary surgeon and approved by the local interdisciplinary tumor board in all cases. The specific tumor anatomy was assessed by endoscopic retrograde cholangiopancreatography (ERCP) or magnetic resonance cholangiopancreatography (MRCP), while the vascular anatomy at the liver hilum was assessed by multiphase computed tomography (CT). Our preoperative work-up included unilateral stenting strategy as standard of care to relieve the future liver remnant (FLR) from cholestasis and bilateral stenting in cases with persisting cholangitis. Endoscopic biliary drainage (EBD) was generally preferred over percutaneous biliary drainage (PBD). In patients with insufficient FLR (< 40%) scheduled for right-sided hepatectomy, a right portal vein embolization (PVE) was conducted 2-4 weeks before surgery. Prior to skin incision, the patients received a single shot antibiotic prophylaxis with cefuroxime and metronidazole in accordance with common guidelines for the prevention of surgical site infections 16 .
The surgical procedure was carried out as previously described by Neuhaus et al. 4,6,9,10,17 . Briefly, a "no-touch" hilar en-bloc resection approach, as defined by liver resection with mandatory portal vein reconstruction, was carried out in all cases and additional arterial resection or the concomitant resection of the pancreatic head (hepatoduodenopancreatectomy) on demand ( Table 1). The actual parenchymal transection was carried out using the Cavitron Ultrasonic Surgical Aspirator (CUSA) with low central venous pressure. The extrahepatic bile ducts and the bile duct bifurcation is removed en-bloc with the liver specimen during parenchymal transection exposing the bile ducts to the FLR at the resection surface. A swab sample of the bile ducts for microbiological testing is immediately taken by the attending surgeon. Afterwards, biliary reconstruction was achieved by an end-to-side hepaticojejunostomy comprising an open hand-sewn anastomosis technique followed by a Rouxen-Y reconstruction. All surgical specimens were evaluated by an experienced board-certified staff pathologist regarding pathological characteristics according to current national guidelines, WHO-and UICC-classifications.
Antibiotic susceptibility of preoperatively colonizing bacteria. Finally, we compared the antibiotic susceptibility of the bacterial bile duct isolates with the standard perioperative antibiotic regimen. While only 12.5% (10/80) of the bile duct colonizing bacterial isolates were susceptible to the standard perioperative antibiotic agent cefuroxime, 75.0% (60/80) were susceptible to a combination of ceftriaxone and vancomycin (Table 5).

Discussion
Radical surgery is the therapeutic mainstay in pCCA as it displays superior oncological outcome compared to palliative treatment modalities 7 . Therefore, improving the perioperative outcome of patients undergoing major hepatic resection for pCCA would significantly improve their overall outcome 1,2,6,8-10 . While preoperative cholangitis has previously been identified as an independent predictor of an adverse outcome, the incidence of bacterial bile duct colonization and its impact on the clinical management and outcome have not been determined 13 In the present study, we investigated the influence of bacterial bile duct colonization on the clinical outcome of patients with pCCA. Here we provide the results of a relatively large European cohort of pCCA patients and demonstrated that bacterial bile duct colonization occurs in the vast majority of patients. We also identified reduced susceptibility of the bacterial isolates to the administered perioperative antibiotic prophylaxis as an independent predictor of postoperative abdominal infections in the multivariate analysis. Based on these observations we speculate that a change in the perioperative antibiotic prophylaxis might significantly reduce postoperative abdominal infections and subsequently improve the overall clinical outcome in pCCA patients.
Bacterial cholangitis has been identified as major risk factor in patients with biliary tract disease such as malign and benign bile duct stenosis or choledocholithiasis among others [18][19][20][21] . However, the main research focus has been the choice of the empirical antibiotic therapy with respect to the common bacterial spectrum of acute cholangitis and not the exploration of bacterial bile duct colonization and its impact on the subsequent surgical intervention. Also, in comparison to our present analysis of pCCA patients, most previous reports have included a rather heterogeneous patient cohort including a broad range of different biliary tract diseases and clinical conditions [18][19][20][21] . The microbiological findings of our patients are in line with these heterogeneous cohorts with Enterococcus spp. being the most relevant gram-positive bacterium and Enterobacter cloacae, Escherichia coli and Klebsiella pneumoniae representing the most common gram-negative bacterial biliary isolates [18][19][20][21] . Interestingly, gram-negative bacteria appeared clinically more relevant in these previous reports as they showed a higher likelihood to be also detected in blood cultures 18,20 . In our patient cohort, gram-positive bacteria, particularly Enterococcus faecium, seemed to have a major clinical importance since preoperative colonization with Enterococcus faecium was associated with a significantly increased risk of postoperative abdominal infection in our univariate analysis (Table 4). Also, gram-positive bacteria were more commonly detected as causative agents of postoperative abdominal infections in these patients with Enterococcus faecium being isolated in 71.2% of all abdominal infections compared to Enterobacter cloacae as the most frequent gram-negative bacterium in only 25.0% of all abdominal infections (Table 1). These results demonstrate notable differences in the clinical significance of bacterial isolates in cholangitis in general and postoperative abdominal infections after surgical resection of pCCA in particular.
Of note, the only available comparable studies to our report, though on the basis of Asian patients, heterogeneous tumor entities and significantly different clinical management and objectives, have been published from the Nagoya group. Sugawara et al. have shown in a randomized-controlled that a two-day and four-day postoperative administration of antimicrobial prophylaxis results in the same clinical outcome 22 . As a secondary finding in this study, preoperative bile cultures were colonized mostly with Enterococcus species, followed by Klebsiella species and Enterobacter species which was similar to our findings and does further underline the importance of gram-positive bacteria in pCCA 22 . In another retrospective cohort study focusing on the role of preoperative biliary drainage, the same group observed an analogous microbial spectrum and indicated that postoperative infections might display the same bacteria as previously detected in preoperative bile samples 23 . Interestingly, preoperative biliary colonization was observed in 84% of all patients comparable with our data, but was also associated with an increased risk of postoperative infections which was not the case in our study 23 . However, the interesting findings of both studies are not transferable to our setting as they included a significant proportion of patients with gallbladder cancer, other malignancies and benign strictures, focused on preoperative bile samples instead of intraoperatively obtained swap samples and most importantly, conducted a different perioperative strategy regarding antibiotics among other differences in clinical management (e.g. use of nasobiliary drainage etc.) 22,23 . In the large retrospective cohort study of Sugawara et al., postoperative antibiotic prophylaxis was used for at least 3 days after operation, even in cases without preoperative bacterial colonization, while this approach was not conducted in our cohort 23 . Further, susceptibility of the cultivated bile samples to the intraoperatively applied antibiotics was not reported in this study 23 .
As already mentioned above, the vast majority of our patients presented with bacterial colonization of the bile ducts at the time of surgery ( Table 1). The most likely explanation for this is the high number of endoscopic interventions in pCCA patients prior to definitive surgery to achieve biliary decompression of the FLR 11,12 . Biliary endoprostheses are a known risk factor for bacterial cholangitis and notability associated with a higher risk of Enterococcus faecium colonization 18,24,25 . This observation was confirmed in our analysis as a biliary plastic stent in situ placed by EBD was significantly more common in patients with bacterial colonization compared to patients without bacterial colonization at the time of surgery. However, we were not able to identify other differences in clinical characteristics between patients with and without bacterial colonization. In particular, while preoperative cholangitis showed a tendency to be more common in patients with bile duct colonization, we observed no difference in the preoperative antibiotic use between individuals with and without bacterial colonization.
Major surgery required for radical resection of pCCA is often associated with significant morbidity and increased perioperative mortality 26,27 . In our particular cohort the 30-day mortality was 14.7% and major Table 2. Group comparison between patients with and without bacterial bile duct colonization. The various agents described here as "Preoperative antibiotics" describe the preoperative administration and do not refer to perioperative antimicrobial prophylaxis. Data presented as median and interquartile range if not noted otherwise. Bold indicates statistical significance (p < 0.05). ASA American society of anesthesiologists classification, BMI body mass index, EBD endoscopic biliary drainage, PBD percutaneous biliary drainage, PVE portal vein embolization. www.nature.com/scientificreports/ morbidity defined as complications categorized ≥ IIIb according to the Clavien-Dindo scale were observed in more than half of the patients. Our comparative analysis of patients with and without postoperative abdominal infections showed significantly more major complications and a tendency for a higher mortality (Table 3). This observation confirms the importance of postoperative abdominal infections as a major cause for an adverse outcome. Subsequently, we conducted a logistic regression analysis to identify independent predictors for postoperative abdominal infection and identified reduced susceptibility of the bacterial bile duct isolates to the perioperatively applied antibiotics as the most relevant predictor in our multivariable analysis (Table 4). During surgery for pCCA, the extrahepatic bile ducts are resected together with the hemi-liver to obtain clear tumor margins 6,[8][9][10] . The biliary reconstruction is afterwards achieved by a hepaticojejunostomy 28 . Intraoperatively the abdominal cavity is inevitably contaminated by biliary secretion from the exposed major bile ducts of the resection surface. The results of our multivariable analysis support the hypothesis of a relevant association between the abovementioned intraoperative exposure of the abdominal situs to bile fluid and postoperative abdominal infections. They may therefore have important implications for the selection of an adequate perioperative antibiotic prophylaxis. This is consistent with the elevated rate of persistence of Enterococcus faecium illustrated by the fact that Table 3. Group comparison between patients with and without postoperative abdominal infection. The various agents described here as "Preoperative antibiotics" describe the preoperative administration and do not refer to perioperative antimicrobial prophylaxis. Data presented as median and interquartile range if not noted otherwise. Bold indicates statistical significance (p < 0.05). ASA American society of anesthesiologists classification, BMI body mass index, EBD endoscopic biliary drainage, PBD percutaneous biliary drainage, PVE portal vein embolization.  Table 6). The close association of bile duct colonization at the timepoint of surgery with postoperative infections underlines the importance of intraoperative bile cultures to better adjust individual postoperative antibiotic treatment and to collect a larger amount of clinical data and eventually optimize future empirical prophylaxis on a larger populational level. Perioperative antibiotic prophylaxis is an internationally accepted measure to reduce surgical site infections in patients undergoing abdominal surgery 16 . Our in-house policy during the study period comprised a single shot antibiotic prophylaxis with a second generation cephalosporine (cefuroxime) plus metronidazole in patients with pCCA in accordance with clinical standards 29 . Of note, only a small subset (12.5%) of the bile duct isolates was susceptible to the antibiotic prophylactic regimen. This reflects the fact that this antibiotic regimen has been selected to prevent surgical site infections in a general surgical population and not to address the specific and inevitable bile contamination of the abdominal cavity during pCCA surgery ( Table 5). As the perioperatively administered antibiotics were the strongest predictor for postoperative abdominal infection in our cohort, we assessed the susceptibility of the bile duct isolates to various additional antibiotics. Here, vancomycin with 43.8% and meropenem with 68.8% were the most effective antibiotics with the combination of both covering 97.5% of all isolates (Table 5).
Our findings identify an important clinical factor with a potentially significant impact in the management of these high-risk patients and suggest a modification of the perioperatively applied antibiotics to improve postoperative outcome. While a combination of meropenem and vancomycin appears to cover the vast majority of all bacterial bile duct isolates, it has to be acknowledged that the application of two broad-spectrum antibiotics during surgery might facilitate the development of multidrug-resistant bacteria and can complicate the selection of antibiotics in patients presenting with septic complications in the early postoperative period. Considering this potential relevant adverse effect, our standard antibiotic prophylaxis was changed to a third generation cephalosporine (ceftriaxone) supplemented by vancomycin to cover Enterococcus faecium (observed in 80.0% of all bile duct isolates).
The duration of perioperative prophylaxis in pCCA is topic of an ongoing debate and typical regimes vary from a classical one-time prophylaxis during surgery to prolonged perioperative prophylaxis including the early postoperative period (e.g. 48 h) 22 . While there is clinical evidence from a randomized-controlled trial of the Nagoya group that there is no difference in outcome between a two-day and four-day postoperative prophylaxis, no systemic evaluation of a one-time prophylaxis versus a short (48 h) postoperative prophylaxis is currently available in the literature 22 . Unfortunately, this relevant question cannot be further explored with the help of our data as our strategy comprised a one-time prophylaxis during the complete study period. However, given the importance of the issue, clinical trials are warranted which should investigate the optimal duration of prophylaxis in this complex disease which may help to further improve surgical outcomes in pCCA patients.
Penetration of the antibiotic agent to the biliary lumen has historically been considered one of the major factors in the selection of antibiotics. Unfortunately, vancomycin's ability to reach the bile fluid (as indicated by a relatively low bile to serum concentration ratio) is limited 30 . However, the current Tokyo Guidelines on antibacterial treatment for cholangitis and cholecystitis still recommend the use of vancomycin in case of Enterococcus faecium associated infections 31 . Moreover, there is some evidence that the secretion of antibiotics into bile fluid is altered by biliary obstruction, a condition frequently observed in pCCA patients despite preoperative drainage 31,32 . Also, the establishment of antibiotic concentrations within the abdominal cavity to reduce the spread of bile-derived bacteria to the abdominal lumen during the intervention might prevent abdominal infection even in the absence of a reduced bile duct colonization.
Similar to other retrospective clinical outcome studies, our analysis has a number of limitations, which should be considered. All patients of this study underwent surgery in a single center in accordance with the authors www.nature.com/scientificreports/ individual approach to pCCA and all data were obtained in a retrospective fashion. Although a multicenter approach may have increased the value of our analysis in terms of sample size, it may also introduce an increased heterogeneity in the clinical management and operative techniques. Unfortunately, as described in the methods www.nature.com/scientificreports/ section, intraoperative bile cultures were not available for all patients who underwent surgery for pCCA during the study period, which may have introduced a certain selection bias. This study represents, to the best of our knowledge, the first European report investigating bacterial bile duct colonization and its influence on the postoperative outcome of patients undergoing curative-intent surgery for pCCA. Despite the aforementioned limitations, our analysis identified bacterial bile duct colonization as a common condition of patients undergoing surgery for pCCA and the reduced susceptibility to the intraoperatively administered antibiotic prophylaxis as an independent predictor for postoperative abdominal infection. As postoperative abdominal infections are a major cause of postoperative morbidity and mortality, a modified perioperative antibiotic prophylaxis might improve the operative outcome in patients with pCCA. Larger multicentric analyses and prospective controlled trials are needed to confirm and validate our findings.

Data availability
Available upon request. JB and UPN had full access to the data and act both as guarantor for the data.