Complications of percutaneous transhepatic cholangiography and biliary drainage, a multicenter observational study

Biliary tract occlusions occur when the biliary system is clogged from within by bile stones or sludge, or when the bile ducts are narrowed by a stricture or due to compression by an external mass. Stasis of bile flow can manifest clinically as jaundice and fatty stools. Furthermore, bacterial colonization in the obstructed biliary system can lead to infections such as cholangitis. Endoscopic Retrograde Cholangiopancreatography (ERCP) is the current gold standard for obtaining biliary access to treat such biliary tract disorders [1, 2]. A drain or stent is placed to restore bile flow from the liver to the gastrointestinal (GI) tract and to keep the common bile duct open. Cannulation rates for ERCP are > 90%, but series have shown failure rates in up to 25%, mostly due to surgically altered anatomy, gastric outlet obstruction, a periampullary diverticulum, an indwelling duodenal stent, or a large tumor [3‐7].

In case of cannulation failure of ERCP, in general three alternative strategies are available to access the biliary tree. The first is a repeat ERCP after 3 days by an expert endoscopist, which is successful in 63–78% of cases [8]. Percutaneous transhepatic biliary drainage [PTC (D)] is another procedure for biliary tree drainage resulting in biliary access in 63–86% of cases [1‐3]. More recently, endoscopic ultrasound-guided biliary drainage (EUS-BD) was introduced as an alternative to PTCD in selected patient groups.

A German single center study also reported a complication rate of 40% in a cohort of 385 PTCD patients [5]. Furthermore, a high mortality rate was reported, i.e., 23% of patients with a benign disorder died after a median of 192 days after the procedure, while those with malignant obstruction had a mortality rate of 70% after a median of 58 days [5].

These studies have raised concerns about the safety of PTCD. A direct translation of abovementioned data to an average ERCP-failure population is difficult, because the reported studies are derived from mostly small, retrospective, and monocenter studies, performed in subgroups of patients. In addition, risk factors associated with complications and reinterventions as a result of PTCD remain to be mapped. Therefore, we aimed to evaluate safety and associated reintervention rates of PTCD in a real-world population obtained from multiple centers.

A total of 331 patients, who underwent PTCD between 2011 and 2016 in 5 centers, one university hospital and four teaching hospitals in the Netherlands, were evaluated. All PTCD were performed after a failed ERCP as a rescue intervention. This study was approved by the medical ethical committees of all participating hospitals (reference number: 2016-0862).

Inclusion criteria were age ≥ 18 years old, one or more successful PTCDs, and at least 1-month follow-up (as registered in the patient’s medical file). Patients with pre-existing infections or fever at the time of the PTCD, or who were currently still receiving antibiotics for an infection in the previous 2 weeks were separately included for assessment of non-infectious complications (for definitions see below).

Primary endpoints were complications and mortality rates within 30 days after PTCD. Complications were divided into non-infectious (bleeding, bile leakage, and stent dislocations) and infectious complications (cholangitis, sepsis, abscess, and cholecystitis).

Secondary endpoints were number of re-interventions during the follow-up time, including anticipated re-interventions such as stent placement, risk factors, and protective factors for complications. The following risk factors were predefined: malignant origin of biliary obstruction, prophylactic antibiotics or not, direct duodenal cannulation, age, BMI > 30, diabetes, bilateral drainage, and whether duodenal cannulation was obtained. After data collection was completed, a random sample of 10% of the data was double checked and uncertainties were resolved through discussion with the involved physicians (EVG, JM, MR, LY, MVS).

The overall incidence of complications and mortality after PTCD is substantial. More than 50% of patients in our cohort developed one or more drainage-related complications after PTCD, mainly cholangitis and sepsis. Mortality seemed to be a direct or indirect result of the PTCD procedure in 27 patients. It is important to keep in mind that all PTCD procedures were performed after a failed ERCP as a rescue intervention. Therefore, these complications may also be the consequence of the previously performed ERCP which may have played a role as well in the significant 30-day morbidity and mortality rates. The majority of these patients had malignant biliary obstruction, further increasing the baseline risk of infectious complications after PTCD. Nonetheless, our study is not the first study that reports complication rates after PTCD that are higher than the threshold as stated in recent SIR guidelines [9]. In 2016, another Dutch RCT comparing the efficacy and safety of pre-operative drainage by ERCP versus PTCD in perihilar cholangiocarcinoma was terminated prematurely because of increased overall mortality rates in the PTCD group (3/27 (11%) vs. 11/27 (41%), respectively) [7]. Cholangitis occurred in 59% in the PTCD-group vs. 37% in the ERCP-group. These findings are in line with our results and clearly shows that infectious complications are common in patients with malignant biliary obstruction, are often drainage-related and seem to occur more often after PTCD than ERCP.

The all-cause mortality rate in our cohort was comparable to earlier studies with 30-day mortality rates of 10–23.1% [6, 10‐12]. We found a high proportion (40.6%) of PTCD-related infectious complications. Previous studies have reported lower infectious complication rates of up to 17% [56, 12‐17]. A possible explanation for these substantial morbidity rates could be the clinical definition of our endpoints. For example, the clinical diagnosis of sepsis following PTCD did not depend on a confirmative blood culture, because in clinical practice blood cultures were often not drawn or only obtained after the start of antibiotic therapy. Another explanation could be the high proportion of patients with a malignant biliary obstruction in our study, a known risk factor for cholangitis and sepsis [11, 18]. Patients with a malignancy usually have a poorer performance status at baseline compared to patients with a benign obstruction [15]. Nonetheless, we did not identify malignant disease as a risk factor for infectious complications. The number of observations of patients with benign obstructions in our study may well have been too low for a robust comparison between these groups.

After the procedure, the majority of patients (78.6%) in our cohort had internalized drainage. This is desirable as it decreases the risk of metabolic disturbances and likely increases patients’ quality of life [19]. Nonetheless, our results show that internalized drainage was an independent risk factor for infectious complications. This seemed to be related to an increase in catheter obstructions, resulting in biliary stasis and cholangitis. These findings are in line with a study that compared complication rates between internal–external drains with a capped-off external component, and external drainage. The authors reasoned that translocation of intestinal flora may have caused (infectious) complications, due to a retrograde flow of intestinal content through the internalized drain [20].

Severity grading is a tool that can potentially discriminate between non severe, moderate severe, and severe complications, and is of utmost importance for future research and quality control in PTCD. Several scores can be used to classify the severity of PTCD complications, for example the ASGE ERCP severity grading and the Clavien-Dindo post-operative complication severity grading [21]. Unfortunately, these scores could not be calculated, because of the incomplete data present in the electronic medical record.

In our study, antibiotic prophylaxis did not prevent infectious complications after PTCD in patients with malignant obstruction. This could possibly be explained by varying antibiotic prophylaxis practices in the Netherlands; we noted a wide variation in timing, type, dosing, and duration of antibiotic prophylaxis. According to the literature, antibiotic prophylaxis prior to PTCD decreases cholangitis risk from 24–46% to 4.6%. [15, 22] Therefore, the 2010 Cardiovascular Interventional Radiological Society of Europe and the Canadian Interventional Radiology Association guideline, advocate antibiotic prophylaxis in PTCD procedures based on level 3–5 evidence [3, 22, 23]. Notwithstanding this, there is currently no standard protocol for antibiotic prophylaxis in PTCD in the Netherlands. Based on the high rate of infectious complications in our study and guideline recommendations, we therefore suggest prescribing pre-interventional intravenous antibiotics to all patients.

This study comes with strengths and limitations. The multicenter design is a major strength of our study. In this large heterogeneous real-life cohort, we report both infectious and non-infectious complications and their associations. Due to the pragmatic definition of endpoints, adjusted to the existing clinical documentation, the results closely mirror everyday clinical practice. Additionally, risk factors for complications and for PTCD-related mortality were analyzed. A strength in our analysis is that we considered the trend of drain obstruction preceding infectious complications in time. Therefore, we were able to confirm that drain obstruction due to sludge or dislocation is an independent risk factor for infectious complications. Furthermore, the multilevel analysis takes the multicenter character of our cohort into account, correcting the OR for any differences between patients from different sites.

Nevertheless, we recognize that the retrospective design of this study has its limitations. The study design is more prone to bias as subjects were not randomized and data collection was dependent on availability and accuracy of registration and follow-up information. Antibiotic prophylaxis strategies differ between hospitals and detailed data such as time of drug administration were often not retrievable. Also, the exact time relation between PTCD and subsequent complications was not structurally documented in the case record form. These factors limit the distinction between early and late complications and additional analysis of the effect of antibiotic prophylaxis in these two groups (See Supplementary Table 2 and Fig. 2). Future studies should focus on clinically relevant definitions of complications, severity grading of complications, quality of life, and comparison of PTCD with advanced endoscopic techniques, such as EUS-guided biliary drainage. Recently, a meta-analysis reported a significant superior performance of EUS-guided biliary drainage compared to PTCD in patients with a malignant obstruction in the common bile duct in terms of technical success (92% vs. 86%, respectively), complication rates (16% vs 80%, respectively), and reintervention rates (16% vs 45%, respectively) [4].

In conclusion, this study shows that PTCD following unsuccessful ERCP is associated with a high number of adverse events, in particular cholangitis and sepsis. The high complication rates justify quality control of the PTCD procedure and pre-procedural prophylactic antibiotic treatment.