Endoscopic ultrasound-guided hepaticogastrostomy and endoscopic retrograde cholangiopancreatography-guided biliary drainage for distal malignant biliary obstruction due to pancreatic cancer with asymptomatic duodenal invasion: a retrospective, single-center study in Japan
Article information
Abstract
Background/Aims
Duodenal invasion (DI) is a risk factor for early recurrent biliary obstruction (RBO) in endoscopic retrograde cholangiopancreatography-guided biliary drainage (ERCP-BD). Endoscopic ultrasound-guided hepaticogastrostomy (EUS-HGS) may reduce early RBO in cases of asymptomatic DI, even when ERCP is possible.
Methods
We enrolled 56 patients with pancreatic cancer and asymptomatic DI who underwent EUS-HGS (n=25) or ERCP-BD (n=31). Technical and clinical success, early (<3 months) and overall RBO rates, time to RBO (TRBO), and adverse events were compared between the EUS-HGS and ERCP-BD groups. Risk factors for early RBO were also evaluated.
Results
Baseline characteristics were similar between the groups. Both procedures demonstrated 100% technical and clinical success rates, with a similar incidence of adverse events (48% vs. 39%, p=0.59). While the median TRBO was comparable (5.7 vs. 8.8 months, p=0.60), EUS-HGS was associated with a lower incidence of early RBO compared to ERCP-BD (8% vs. 29%, p=0.09). The major causes of early RBO in ERCP-BD were sludge and food impaction, rarely occurring in EUS-HGS. EUS-HGS was potentially reduced early RBO (odds ratio, 0.32; p=0.07).
Conclusions
EUS-HGS can be a viable option for treating pancreatic cancer with asymptomatic DI.
INTRODUCTION
Malignant biliary obstruction (MBO) is a common clinical manifestation of pancreatic cancer. Currently, endoscopic transpapillary self-expandable metal stent (SEMS) placement via endoscopic retrograde cholangiopancreatography (ERCP) is the mainstay of biliary drainage for distal MBO.1 However, in cases of duodenal invasion (DI), transpapillary-placed SEMS can be susceptible to early occlusion by sludge formation and food impaction due to increased duodenobiliary reflux.2,3 As DI occurs in up to one-third of patients with advanced pancreatic cancer and is associated with early SEMS dysfunction,3 endoscopic ultrasound-guided biliary drainage (EUS-BD) may be considered as an alternative to ERCP-guided biliary drainage (ERCP-BD) in this population.
EUS-BD has emerged as an alternative to percutaneous transhepatic biliary drainage after failed or challenging ERCP, especially in cases of gastric outlet obstruction (GOO) or surgically altered anatomy.4-6 With accumulating evidence of EUS-BD as a salvage treatment, it is now recognized as a treatment option for biliary drainage in patients with MBO, even if ERCP is technically possible.7-10 We previously reported that EUS-BD offers longer patency than ERCP-BD in patients with an indwelling duodenal stent.11 Furthermore, a recent study suggested that EUS-guided hepaticogastrostomy (EUS-HGS) provides longer stent patency with fewer adverse events (AEs) than EUS-guided choledochoduodenostomy in this situation.12 Based on these study results, EUS-HGS is increasingly used in combination with MBO and GOO in clinical practice. However, it remains unclear whether EUS-HGS can provide better clinical outcomes in cases of asymptomatic DI, where ERCP-BD is technically possible but prone to duodenobiliary reflux.
In this retrospective study, we aimed to assess the safety and effectiveness of EUS-HGS and ERCP-BD for treating distal MBO caused by pancreatic cancer with asymptomatic DI.
METHODS
Patients
We retrospectively examined patients who underwent either EUS-HGS or ERCP-BD using SEMS for distal MBO complicated with asymptomatic DI due to unresectable pancreatic cancer at our institution between January 2010 and March 2022. EUS-HGS or ERCP-BD were primarily selected on a chronological basis, with a preference for EUS-HGS since 2020. As shown in Figure 1, DI was diagnosed based on tumor infiltration or abutment to the duodenum observed on computed tomography (CT), coupled with the corresponding findings of duodenal erosions, ulcers, or strictures on endoscopy due to pancreatic cancer, irrespective of pathological confirmation. Patients with symptomatic GOO requiring gastroduodenal stent placement were excluded because ERCP-BD is often technically impossible. Patients with a history of biliary drainage, excluding temporary transpapillary plastic stent placement or nasobiliary drainage within 1 month before SEMS placement, were excluded from the study. Additionally, patients with a surgically altered anatomy, except for those who underwent Billroth I reconstruction, were excluded.
Endoscopic procedure
The details of EUS-HGS and ERCP-BD have been reported elsewhere.13,14 A partially covered SEMS (modified Giobor and Spring Stopper stents; Taewoong Medical) was placed from the left intrahepatic bile duct (B2 or B3) to the stomach, with the uncovered portion located in the bile duct. The SEMSs used for EUS-HGS were 8 or 10 mm in diameter and 10 or 12 cm in length. In this study, the SEMSs used for ERCP-BD were fully or partially covered, with a diameter of 10 or 12 mm and a length of 4 to 8 cm. The selection of SEMS for each procedure was at the discretion of the attending endoscopists.
Outcomes and definitions
The primary outcome of this study was recurrent biliary obstruction (RBO), including early RBO. The secondary outcomes included technical success, clinical success, reinterventions for RBO, AEs, and overall survival (OS).
Technical success, clinical success, RBO, and AEs were defined according to the Tokyo criteria.15 Technical success was defined as a successful stent deployment at the intended location. Clinical success was defined as a 50% decrease in or normalization of the bilirubin level within 14 days of stent placement. RBO was defined as a composite endpoint of either occlusion or migration, and early RBO was defined as RBO occurring within 3 months after the procedure. Time to RBO (TRBO) referred to the time from SEMS placement to RBO. AEs were classified as early (within 14 days) or late (after 14 days) and graded according to the severity grading system of the American Society for Gastrointestinal Endoscopy lexicon.16 Peritonitis was defined as the presence of clinical symptoms of peritoneal irritation and the corresponding accumulation of fluid on CT. Absence of fluid collection on CT despite the presence of abdominal pain was considered as abdominal pain and not peritonitis.14 OS was defined as the time from SEMS placement to final follow-up or until death from any cause.
Statistical analysis
The results are expressed as numbers (%) or medians with ranges. Categorical and continuous variables were compared using the Fisher exact test and Student t-test, respectively. Cumulative TRBO and OS were calculated using Kaplan-Meier analysis and compared using a log-rank test. Deaths without RBO were censored at the time of death. The risk factors for early RBO were evaluated using logistic regression analysis, which included the following variables: age, sex, performance status score, tumor size, liver metastasis, ascites, EUS-HGS, gastroduodenal stent, and chemotherapy after biliary SEMS placement. To prevent overfitting due to the small number of cases with early RBO, only factors with p-value ≤0.10 in the univariate analyses were included in the multivariate analysis. Statistical significance was set at values of p<0.05. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University), a graphical user interface for R (The R Foundation for Statistical Computing).17
Ethical statements
This study was approved by the Ethics Committee of The University of Tokyo Hospital (2018171NI) and performed in accordance with the Declaration of Helsinki. The requirement for written informed consent was waived by the Ethics Committee because of its retrospective design. A notification of this study was published on our hospital website, allowing patients to opt out of the study.
RESULTS
Patients
A total of 282 patients were diagnosed with unresectable pancreatic cancer complicated with distal biliary obstruction, of whom 56 patients with asymptomatic DI were finally included in this study (25 for EUS-HGS and 31 for ERCP-BD, Fig. 2). The remaining 226 patients were excluded, including those who underwent biliary drainage prior to referral (n=19), those with a history of plastic stent placement for more than 1 month (n=12), those with surgically altered anatomy except for Billroth 1 reconstruction (n=7), those without DI (n=168), and those with symptomatic GOO, or those who underwent gastroduodenal stent placement prior to biliary drainage (n=20).
Patient characteristics are summarized in Table 1. No significant difference was observed in age and sex distribution, performance status score, disease status (i.e., primary tumor size, liver metastasis, and ascites), or DI location between the two groups. Furthermore, the rates of gastroduodenal SEMS placement for GOO and chemotherapy after biliary SEMS placement were comparable between the groups (24.0% vs. 16.1%, p=0.51 and 60.0% vs. 74.2%, p=0.39, respectively). As shown in Figure 3A, the median OS was 3.2 (95% confidence interval [CI], 2.5–6.7) and 5.9 (95% CI, 3.5–9.9) months in the EUS-HGS and ERCP-BD groups, respectively (p=0.60).

(A) Kaplan-Meier curves for overall survival in patients with endoscopic ultrasound-guided hapaticogastrostomy (EUS-HGS, solid line) and endoscopic retrograde cholangiopancreatography (ERCP-BD, broken line). (B) Kaplan-Meiers curve for time to recurrent biliary obstruction in patients with EUS-HGS (solid line) and ERCP-BD (broken line).
Technical success, clinical success, and AEs
The details of the procedures are shown in Table 2. EUS-HGS and ERCP-BD were technically successful in all cases. The median procedure times were 30 minutes (range, 18–77 minutes) and 40 minutes (range, 12–78 minutes) for EUS-HGS and ERCP-BD, respectively (p=0.25). The stent diameter used in EUS-HGS was 8 mm and 10 mm in 10 and 15 patients, respectively, whereas in ERCP-BD, it was 10 mm and 12 mm in 27 and 4 patients, respectively. Clinical success was achieved in all EUS-HGS and ERCP-BD cases.
The details of AEs other than RBO are presented in Table 3. The incidence of overall AEs was comparable between the EUS-HGS and ERCP-BD groups (48.0% vs. 38.7%, p=0.59), but their profiles differed, especially for early AE. Common early AEs included transient fever (12.0%), abdominal pain (12.0%), and peritonitis (8.0%) in EUS-HGS and cholecystitis (12.9%) and pancreatitis (6.5%) in ERCP-BD. Regarding late AEs, liver abscess and cholecystitis were observed in both groups, whereas pseudoaneurysm was noted in the ERCP-BD group. All AEs were successfully managed with conservative treatment and no procedure-related deaths were observed in either group.
RBO
The overall and early RBO are listed in Table 4. Within a median follow-up duration of 3.1 months for EUS-HGS and 2.9 months for ERCP-BD, RBO occurred in 8 (32.0%) and 15 (48.4%) patients treated with EUS-HGS and ERCP-BD, respectively (p=0.28). The major causes of RBO were hyperplasia in the uncovered portion of the SEMS in the EUS-HGS group (24.0%) and sludge, food impaction, and migration in the ERCP-BD group (12.9%, 9.7%, and 9.7%, respectively). As shown in Figure 3B, the median cumulative TRBO was 5.7 (95% CI, 3.2–not reached [NR]) and 8.8 (95% CI, 2.8–NR) months in the EUS-HGS and ERCP-BD groups (p=0.60). The non-RBO rates at 1, 3, and 6 months were 92%, 52%, and 4%, respectively, in the EUS-HGS group, and 71%, 39%, and 23%, respectively, in the ERCP-BD group. Although the cumulative TRBO and incidence of overall RBO were comparable between the groups, the incidence of early RBO was lower in the EUS-HGS group than in the ERCP-BD group (8% vs. 29%, p=0.09). The major contributors to early RBO in the ERCP-BD group were identified as sludge (13%) and food impaction (10%), both of which rarely occurred in the EUS-HGS group.
Table 5 shows risk factors for early RBO identified through univariate and multivariate logistic regression analyses. In the multivariate analysis, only EUS-HGS was associated with low early RBO (odds ratio, 0.32; 95% CI, 0.12–1.08; p=0.07).
Reintervention for RBO
In all eight patients with RBO after EUS-HGS, reintervention was successfully performed using the existing EUS-HGS route. Among six patients with RBO due to hyperplasia, stent-in-stent placement was performed using an uncovered SEMS (n=4) or a plastic stent (n=2). For two patients with RBO caused by sludge, a balloon sweep through the HGS route was performed, and an additional SEMS was placed across the papilla in one patient. Conversely, 2 of 15 patients with RBO after ERCP-BD required conversion to EUS-HGS because transpapillary reintervention was technically difficult due to an indwelling gastroduodenal SEMS for symptomatic GOO. In the remaining 13 cases, new transpapillary stent placement followed by initial SEMS removal (n=10) or stent-in-stent placement with the initial SEMS left in place (n=3) was performed using an SEMS (n=10) or a plastic stent (n=3). Notably, none of the patients required percutaneous biliary drainage as a salvage treatment.
A second RBO occurred in 33% and 38% of the patients after reinterventions through the EUS-HGS and ERCP-BD routes, respectively. The cumulative TRBO of reintervention through the EUS-HGS and ERCP-BD routes was 11.4 months (95% CI, 5.5–NR) and 6.4 months (95% CI, 2.3–NR), respectively (p=0.51).
DISCUSSION
The present study demonstrated that compared with ERCP-BD, EUS-HGS was associated with a lower risk of early RBO (8% vs. 29%, p=0.09), without compromising safety in patients with distal MBO caused by pancreatic cancer with asymptomatic DI. This suggests the potential superiority of EUS-HGS in cases of DI, even when ERCP is technically feasible. However, EUS-HGS did not contribute to improve cumulative TRBO (median TRBO of 5.7 months in EUS-HGS and 8.8 months in ERCP-BD, p=0.60), indicating a need for further investigations to ensure long-term outcomes of EUS-HGS.
We previously reported that DI is a risk factor for early stent dysfunction in ERCP-BD in pancreatic cancer,2,3 and we postulated that EUS-HGS would be theoretically less susceptible to duodenal reflux due to the stagnation of duodenal contents, even in cases with asymptomatic DI.11,18 Our findings revealing a lower incidence of early RBO with EUS-HGS compared to that with ERCP-BD support this hypothesis, although the difference between the procedures is not statistically significant. Although the cumulative incidence of RBO was comparable, the reduced incidence of early RBO in EUS-HGS was clinically relevant in maintaining quality of life and resuming or continuing chemotherapy in this palliative setting, particularly given the limited survival in most cases of advanced pancreatic cancer with DI (median OS was <6 months in the present study).
No significant difference was observed in the overall incidence of RBO or cumulative TRBO between EUS-HGS and ERCP-BD. This can be attributed to the fact that the benefit of avoiding early RBO in EUS-HGS was offset by the high occurrence of RBO resulting from hyperplasia 3 months after SEMS placement. Although the mechanism of hyperplasia formation remains unclear, a mismatch in the diameter between SEMS and the small intrahepatic bile duct or chronic inflammation due to cholangitis may lead to hyperplastic change.14 Thus, a fully covered SEMS or small-diameter SEMS may mitigate the risk of RBO due to hyperplasia19,20; however, it also raises concerns regarding stent migration. Therefore, to further improve the long-term outcomes of EUS-HGS, it is desirable to develop an SEMS that can effectively prevent both hyperplasia and stent migration.
Consistent with previous systematic reviews21,22 the overall AE rates were similar between EUS-HGS and ERCP-BD in the present study (48% vs. 39%, p=0.59), even though the safety profile was substantially different between the groups. Although the incidence of AEs in our ERCP-BD cohort seemed acceptable, ERCP can be technically demanding, especially in cases with DI because of the narrowed and fixed duodenum, which may lead to an increased risk of AEs such as perforation.23-25 Furthermore, in ERCP-BD, pancreatitis can occur due to the occlusion of the pancreatic duct orifice caused by SEMS placement. Additionally, patients without pancreatic parenchymal atrophy are reportedly at a higher risk of post-ERCP pancreatitis.26 Therefore, EUS-HGS may be a good alternative, particularly for patients without pancreatic duct involvement. In contrast, events related to bile leakage and bleeding are common AEs in EUS-HGS, regardless of the presence of DI.14 Thus, the utilization of a dedicated one-step device for tract dilation followed by stent insertion, as well as several techniques such as segmental tract dilation and bile aspiration after bile duct puncture, may decrease the risk of bile leakage.27,28 The rate of procedure-related peritonitis in the present study was lower than that reported in previous studies among patients with MBO who underwent EUS-HGS with SEMS; however, it was as high as one-third in patients with ascites (n=2/6, data not shown). Although EUS-HGS is technically possible for massive ascites,29 ERCP-BD may still be preferred over EUS-HGS in this condition.
This study has some limitations. First, this was a non-randomized, retrospective study with a small sample size, conducted in a single center, thereby carrying a risk of bias. While treatment selection was left to the attending physician’s discretion, various factors such as severe DI could have influenced treatment preferences. Severe DI may introduce technical difficulties in performing ERCP-BD, potentially leading to a bias towards favoring EUS-HGS over ERCP-BD. Second, various SEMSs with different properties were used in ERCP-BD. This could have potentially led to non-uniform outcomes with ERCP-BD, though the aim of this study was to compare the EUS and ERCP approach routes rather than the performance of each SEMS. Finally, the follow-up period was relatively short in this study. However, this was inevitable because of the poor prognosis of the patients with advanced pancreatic cancer with DI.30 To overcome these limitations, we are conducting a prospective randomized trial to compare ERCP-BD and EUS-HGS with SEMS placement in distal MBO with DI (UMIN 000048708).
In conclusion, EUS-HGS demonstrated low incidence of early RBO with comparable safety and can serve as an alternative to ERCP-BD in patients with pancreatic cancer complicated by asymptomatic DI.
Notes
Conflicts of Interest
Yousuke Nakai received research grants from Boston Scientific Japan, Fujifilm Corporation, and HOYA Corporation, and honoraria from Boston Scientific Japan, Fujifilm Corporation, Olympus Corporation, and Gadelius Medical. Mitsuhiro Fujishiro received research grants from Olympus Corporation and Fujifilm Corporation, and honoraria from Fujifilm Corporation and Olympus Corporation. The other authors declare no conflict of interest for this article.
Funding
None.
Author Contributions
Conceptualization: NT, YN, TH; Data curation: NT; Formal analysis: NT, YN; Investigation: all authors; Methodology: NT, YN, TH; Project administration: YN; Resources: all authors; Software: NT; Supervision: MF; Writing–original draft: NT, YN; Writing–review & editing: all authors.