Use of microbiological and patient data for choice of empirical antibiotic therapy in acute cholangitis

Acute cholangitis is a potentially life-threatening infection of the biliary tract. The main underlying pathological process is obstruction of bile ducts leading to cholestasis and impeded clearance of intestinal microorganisms ascending from the duodenum. The increase in biliary tract pressure and the local inflammatory response to the infecting organisms impair the integrity of the biliary epithelium, facilitate bacterial translocation into the circulation, and promote systemic inflammation [1]. In addition, certain procedures, such as sphincterotomy, stent placement or biliodigestive anastomosis, can predispose to bacteriobilia and cholangitis. In these situations, infection of the biliary tract may develop in the absence of obstruction [1, 2].

The clinical course of acute cholangitis ranges from mild abdominal symptoms to sepsis and shock. Despite considerable progress in the treatment of acute cholangitis during the last decades, recent studies still report mortality rates of up to 10% [2, 3]. As with other localized infections, treatment success depends on efficient source control and appropriate antibiotic therapy. Source control is achieved by biliary decompression by endoscopic or percutaneous drainage. The importance of antibiotic therapy is highlighted by the fact that inappropriate empirical therapy is associated with increased mortality [3].

Studies from different geographic regions of the world consistently found E.coli, Klebsiella spp. and Enterococcus spp. as most common pathogens in ascending cholangitis [4]. Thus, the microbial spectrum to be covered by empirical therapy seems to be well defined in general. However, individual patient factors as well as local resistance patterns can increase the probability for infection with resistant enterobacterales, P.aeruginosa or other difficult-to-treat pathogens [5]. Enterobacterales resistant to third generation cephalosporines (3GCRE) are of particular importance because their treatment regularly requires the use of carbapenems. Because of the rising prevalence of cephalosporine resistance, the clinician is confronted with the dilemma of ensuring broad empirical antibiotic coverage while at the same time avoiding excessive use of carbapenems. Particularly in institutions with a high prevalence of 3GCRE, it seems difficult to define a role for carbapenem-sparing regimens. Knowledge of local susceptibility rates alone is not sufficient to solve this problem.

In our study, we investigated whether the analysis of patient factors and their association with pathogens and antibiotic resistance could be helpful to avoid overuse of carbapenems and other broad-spectrum antibiotics. To this end, we conducted a retrospective analysis of 348 patients treated for ascending cholangitis in Germany’s largest university medical center. We determined the pathogen spectrum and antibiotic susceptibilities of enterobacterales and enterococci in biliary and blood samples from these patients. Particularly, we analysed the significance of various patient factors for predicting the isolation of bacterial biliary pathogens and their antibiotic susceptibility profile. Our results demonstrate, how the combination of microbiological and patient data can be used for the choice of efficacious antibiotic therapy of patients with acute cholangitis.

All consecutive specimens sent to the microbiological laboratory of the Charité (Institute of Microbiology and Hygiene of the Charité until December 2010; Labor Berlin Charité Vivantes GmbH, thereafter) from January 2007 until July 2015 were screened for positive bile cultures obtained by endoscopic or percutaneous sampling. From all patients with positive bile cultures, blood culture results were also extracted, if they had been collected within 48 h from the corresponding bile sample.

The diagnosis of cholangitis was reviewed against the patient records. Bile cultures that were not collected within a course of cholangitis were excluded. Samples were also excluded, if the patient’s age was < 18 years or clinical data were incomplete or inconsistent. Figure 1 depicts the schematic structure of the study.

Bile specimens were cultivated on conventional solid media and incubated in aerobic and anaerobic atmospheres at 37 °C. Aerobic microorganisms were routinely identified and tested for antibiotic susceptibility using the VITEK 2 System (bioMérieux, France). Antibiotic susceptibilities were interpreted according to EUCAST recommendations (http://​www.​eucast.​org/​). If necessary, additional commercial methods were applied such as matrix-associated laser desorption/ionization time-of-flight mass spectrometry (MALDI-MS; Bruker, Massachusetts, USA, before 2011, or bioMérieux, France, since 2011) for species identification or the disk diffusion method and E-tests for antibiotic susceptibility testing. Anaerobes were identified by MALDI-MS and tested for susceptibility using ATB ANA (bioMérieux, France) or E-tests. For this study, “intermediate” (I) results in antimicrobial sensitivity testing (AST) were considered “resistant” (R). For enterobacterales, AST was done using ampicillin/sulbactam, piperacillin/tazobactam, cefotaxime, meropenem, and ciprofloxacin. Enterococcus strains were investigated on ampicillin and vancomycin resistance, respectively. Blood cultures with Coagulase-negative Staphylococci (CoNS) in only one pair of bottles were considered contaminated. In line with the recommendations of the German Commission for Hospital Hygiene and Infection Prevention at the Robert Koch Institute (KRINKO) [6] enterobacterales, P. aeruginosa, and A. baumannii complex isolates were classified as multi-resistant gram-negative rods (MRGN) if they were resistant to three or more of the following substance classes: acylureidopenicillins, third generation cephalosporins, carbapenems, fluorochinolones.

Patient data were extracted from the hospital records. In patients hospitalized several times for recurrent cholangitis, only the most recent episode was analysed. Also, in patients with more than one procedure of bile sampling during the same hospital stay, only culture results obtained during the first sampling procedure were considered for correlation of patient factors with microbiological data. Data on empirical antibiotic treatment were available for 229 patients.

The statistical evaluation was carried out after consultation with the Institute for Biometry and Clinical Epidemiology of the Charité - Universitätsmedizin Berlin using SPSS version 23.0. Significant differences in categorical variables were assessed using the Chi-2 test. Predictors with p ≤ 0.2 in the univariate analysis were then tested using multivariate binary logistic regression through backward elimination and forward selection. The significance level in two-sided testing was p < 0.05.

Empirical therapy of cholangitis represents an educated guess based on spectrum and likelihood of causative pathogens and their expected antimicrobial susceptibility. Accordingly, the current Tokyo guideline recommends third generation cephalosporines, piperacillin/tazobactam or carbapenems for empirical treatment the choice of which should be guided by local susceptibility data and the severity of the infection. For severe cases and hospital-acquired infections anti-pseudomonal agents and coverage of enterococci are recommended. Because optimal empirical treatment may vary greatly between different institutions, a multi-disciplinary approach is suggested [4].

Although these recommendations provide a useful framework in general, the selection of antibiotic therapy is still associated with various uncertainties. Without guidance by local susceptibility data, it is difficult for the attending physician to make his choice between the three betalactam-options. Carbapenems seem to be the safest choice, because of their bactericidal activity against most gram-negative rods resistant to third generation cephalosporins and acylureidopenicillins. However, increased carbapenem usage promotes carbapenem-resistance, which is regarded a major global public health problem. According to recognized principles of antibiotic stewardship, it is therefore widely advocated that carbapenems should be used in a restrictive manner in order to minimize the selection pressure favouring growth of carbapenem-resistant bacteria.

Even if local susceptibility rates are available, they are usually not provided in respect to specific disease entities despite considerable disease-specific variations making the choice of the optimal empirical therapy difficult. Owing to the rising prevalence of resistant gram-negative rods, carbapenems often seem to be the only effective option left. Particularly, empirical therapy guided by microbiological data only may fall short in respect to the goals of modern antibiotic therapy if patient-related response rates are not taken into account. Therefore, the present study aimed to analyse both microbiological as well as patient data in order to optimize empirical therapy of acute cholangitis. To this end, our analysis identified risk factors for 3GCRE and/or MRGN. Although we did find a high overall prevalence of cepholosporine resistant enterobacterales in bile and blood cultures of our patients, our results imply that piperacilline/tazobactam is a highly effective option for patients with community-acquired cholangitis, without pre-existing biliary prosthesis, and not treated on an ICU.

In our cohort, polymicrobial growth occurred in almost 70% of bile samples with enterobacterales and enterococci being the most common pathogens, each accounting for about one third of bile isolates. Anaerobes were detected in less than 10% of our patients. These results are consistent with other recent reports [7‐10] and revise the classic view that both enterobacterales and anaerobes are the most important pathogens in biliary tract infections [11]. Bacteremia was detected in one third of cases with available blood culture results. Reflecting their differing virulence, and in agreement with a recent large multicenter trial on biliary infections, enterobacterales comprised almost 60% and enterococci only 20% of blood culture isolates [2]. In line with previous studies, our multivariate analysis found the presence of a biliary prosthesis an independent predictor for enterococcal isolation from bile [3, 9, 12, 13]. The Tokyo guideline recommends empirical treatment covering enterococci in severe community-acquired cholangitis – and in health-care associated infections if patients are colonized or enterococci are otherwise of concern. We agree on this but would like to point out that enterococcal coverage should particularly be considered in patients carrying an indwelling biliary drainage. As stated in the guideline, VRE carrier status should be taken into account in the choice of substance [4].

As to the frequency of resistant enterobacterales, we found an 80% susceptibility of bile isolates to fluorochinolones and an almost 90% susceptibility of blood isolates to third generation cephalosporines. Thus, on first view, all antibiotic choices recommended by the Tokyo guideline seem to be appropriate for our patients. However, when coverage was analysed in a patient-wise manner, a different picture emerged. Although the antibiotics prescribed largely reflected the Tokyo guideline recommendation, empirical therapy covered bile isolates in only 51% of patients. In patients with bacteremic infections, coverage was 69% which is still far less than desired. Probably owing to both the high proportion of polymicrobial growth and the high prevalence of resistant bacteria in bile and blood cultures, the actual antibiotic coverage was unacceptably low.

Epidemiological data from Germany suggest a frequency of 3GCRE of about 15% in hospitals [14]. In an analysis of cholangitis cases from two other German university hospitals, ESBL producing E.coli were detected at a rate of 31% [10]. In our cohort, the highest numbers of cholangitis associated with either 3GCRE or MRGN were found amongst patients carrying MDR pathogens. In those known to be colonized, MRGN were detected in 41% of bile cultures. Individuals with a high burden of comorbidity were also more prone to infections with 3GCRE. Furthermore, the presence of biliary stenosis was associated with an increased risk for MRGN, and we observed a linear trend between the proportion of patients infected with 3GCRE and MRGN and the number of prior biliary tract interventions. Thus, it seems reasonable to cover patients with carrier status, a high burden of comorbitiy or the presence of biliary stenosis and repeated biliary interventions (in our study > 3 interventions) with a carbapenem for empiric therapy. In addition to these risk factors, we found male sex to be an independent risk factor for isolation of 3GCRE or MRGN. In fact, a previous study on acute cholangitis from two German university centers also reported an association of male sex with infections caused by ESBL E.coli [10]. At the time being, the causal relationship for this association is largely unclear. Therefore, we would presently not suggest to cover all male cholangitis patients with a carbapenem.

Having defined predictors for infection with 3GCRE or MRGN, the question arises, whether patient subgroups can be identified that can still be effectively treated with a carbapenem-sparing therapy even in a tertiary referential center with a high prevalence of resistant pathogens. According to our data, piperacillin/tazobactam would represent a highly effective option for patients presenting with community acquired cholangitis, without indewelling biliary drainage, and not on an ICU. Of note, if only resistance data would have been considered (frequency of 3GCRE in blood culture isolates 11.4%), the choice would have been a third generation cephalosporine. However, if analysed in a patient-based manner, the coverage by cefotaxime was poor (< 80%) for unselected patients as well as for all subgroups investigated. The better results for piperacillin/tazobactam in our cohort are explained by the coverage of ampicillin-susceptible enterococci, piperacillin-susceptible P.aeruginosa and anaerobes but not 3GCRE that were tested susceptible to piperacillin/tazobactam in vitro. This is important because piperacillin/tazobactam was inferior to carbapenems in the treatment of blood stream infections caused by E.coli and K.pneumoniae resistant to third generation cephalosporines [15]. Aside from these considerations, our results demonstrate that i) efficacy of antibiotic regimens should be assessed in a patient-based manner, and ii) combined analysis of microbiological and patient data is a valuable tool for deliberate and efficacious use of antibiotics in settings with a high prevalence of resistance.

Since this paper focuses on antibiotic therapy, we have not included a risk analysis for biliary candidiasis. For completeness, a high Charlson Index, bile sampling later than 48 h after admission, and an indwelling drainage were independent risk factors for a fungibilia in our cohort. An indwelling drainage and/or ICU treatment were risk factors for candidaemia .

An important limitation of this retrospective study is the lack of documented information regarding administration of antibiotic therapies prior to microbiological sampling. Owing to its potential effect on microbial spectra and prevelance of resistancies, prior antibiotic exposure should be considered when selecting an empiric regimen. Still, the predictors for a carbapenem-sparing therapy identified in our study can be applied to patients without recent antibiotic therapy and to those in which information on antibiotic exposure cannot be reliably obtained from the patient’s history.

In summary, we found a high prevalence of 3GCRE or even MRGN in bile and blood samples obtained from patients with acute cholangitis as would have been expected for a tertiary referential center with associated liver transplant unit. Carrier status, a high burden of comorbitiy, the presence of biliary stenosis, and repeated biliary tract interventions were identified as risk factors for isolation of resistant gram negative bile culture isolates. Only the combined analysis of microbiological and patient data revealed piperacillin/tazobactam as an excellent choice for empiric therapy of patients with community acquired cholangitis, no biliary prosthesis, and no previous or ongoing ICU treatment. The study thus provides an example of how the Tokyo guideline recommendations can be specified for a particular institution on the basis of local microbiology and individual patient characteristics.