We retrospectively analyzed consecutive patients who underwent BAE-ERCP for benign HJAS at Cancer Institute Hospital of Japanese Foundation for Cancer Research between July 2014 and December 2020. We defined HJAS as a hepaticojejunostomy anastomosis requiring balloon dilation that was associated with cholangitis, jaundice, intrahepatic stones, or elevated liver enzymes. We defined benign HJAS as HJAS without irregular mucosa or rough stenosis on endoscopy, irregular lesions in the bile duct on cholangiography, or findings suggestive of malignancy on computed tomography. Oral anticoagulants were either suspended or replaced with intravenous heparin prior to all procedures. Patients who had previously undergone BAE-ERCP for HJAS and those who required biliary stent deployment in addition to balloon dilation were excluded.

BAE-ERCP was performed using single-balloon enteroscopy under moderate sedation with midazolam and analgesia with pethidine. Two types of single-balloon enteroscopy were used during the study period (SIF-Q260, working length: 2,000 mm, channel diameter: 2.8 mm, and SIF-H290S, working length: 1,520 mm, channel diameter: 3.2 mm; Olympus Medical Systems). All procedures were performed by experts or trainees under their guidance.

After reaching the hepaticojejunostomy anastomotic site, biliary cannulation and cholangiography were performed using a cannulation catheter (MTW Endoskopie Manufaktur, Wesel, Germany) and a 0.025-inch guidewire (Visiglide 2; Olympus Medical Systems). After inserting the guidewire into the bile duct, a catheter was inserted, and cholangiography was performed to measure the diameter of the bile duct above the HJAS. If the HJAS was too narrow for the catheter to pass through, a dilation balloon catheter (CRE PRO or Hurricane RX; Boston Scientific Corp.; REN, KANEKA; MEDIX Corp.) was inserted over the guidewire. If a dilation balloon catheter also failed to pass through the HJAS, a stent retriever (7-Fr screw-type device, Soehendra Stent Retriever; Cook Medical Inc., Bloomington, IN, USA) was used to pre-dilate the HJAS over the guidewire before inserting the balloon catheter.22 The balloon dilator with a balloon diameter of 4 to10 mm (depending on the diameter of the bile duct) was inflated at the HJAS for 30 to 120 seconds (with pressure of up to 810.6 kPa). Balloon dilation was generally conducted only once, although it was repeated up to four times if dilation was deemed insufficient. A residual balloon notch was considered positive if a visible notch remained in the balloon dilator with maximal pressure applied on fluoroscopy immediately before deflation in all dilation attempts. Fluoroscopic images of the absent and present residual balloon notches are presented in Figure 1, respectively.

Any intrahepatic stones were removed using a retrieval balloon catheter and/or basket catheter. An endoscopic nasobiliary drainage tube was placed at the discretion of the physician, generally in patients with concomitant cholangitis or residual intrahepatic stones.

Patients were followed-up with laboratory studies every 1 to 3 months after discharge. BAE-ERCP was repeated in patients with findings suggestive of recurrent stenosis, such as jaundice, fever, elevated liver enzymes, and/or a dilated biliary tree on imaging. Balloon dilation with or without stent deployment was performed to treat recurrent stenosis. Follow-up was conducted until November 30, 2021.

This study evaluated technical and clinical success, HJAS patency after balloon dilation, risk factors for recurrent stenosis, and adverse events (AEs). Technical success was defined as endoscopically visible improvement of benign HJAS after balloon dilation. Clinical success was defined as an improvement in clinical symptoms and/or laboratory data within 14 days after the procedure. The final preoperative computed tomography scan was used to determine the preoperative bile duct diameter. We considered whether the surgery involved vascular resection and whether bile leakage occurred postoperatively. The length of the stenosis was measured from the intestinal wall to the lower end of the dilated bile duct using fluoroscopy. The presence or absence of a scar-like appearance at the anastomosis site on endoscopy was evaluated.23 Patency time was defined as the time from the first treatment to the second balloon dilation or to the last follow-up date. According to the presence or absence of recurrent HJAS, the patients were divided into recurrent HJAS group and no recurrence group. AEs were defined as any postoperative event based on a lexicon for endoscopic AEs.24

Continuous variables are presented as medians (ranges) and compared using the Mann–Whitney U test. Categorical variables are described as absolute numbers (proportions) and were analyzed using the chi-squared or Fisher exact test. A p-value <0.05 was considered statistically significant. Patency time was calculated using the Kaplan–Meier method and compared using the log-rank test. Nine variables were assessed using univariate analysis to identify factors predicting recurrent stenosis as follows: sex, history of hepatectomy, history of preoperative biliary drainage, HJAS occurring within 1 year postoperatively, presence of jaundice, use of a stent retriever, length of stenosis exceeding 3 mm, residual balloon notch, and intrahepatic stone removal. Factors with p-values <0.15 in the univariate analysis were used in the multivariate logistic regression analysis, and hazard ratios (HRs) with a 95% confidence interval (CI) were calculated. All statistical analyses were performed using EZR ver. 1.54 (Saitama Medical Center, Jichi Medical University).25

This study was approved by the ethics committee of Cancer Institute Hospital of Japanese Foundation for Cancer Research (approval number: 2021-GA-1017). The requirement of informed consent for study participation was waived owing to the retrospective nature of the study. All the procedures were performed in accordance with the principles of the Declaration of Helsinki.

Fifty-four patients underwent balloon dilation during the first episode of benign HJAS using BAE-ERCP, of whom two patients were excluded because of ERCP failure. Endoscopic ultrasound (EUS)-guided hepaticogastrostomy was performed in one case because the hepaticojejunostomy anastomosis could not be identified. PTBD was performed in another case because the guidewire could not pass through the heavily stenosed HJAS. The technical success rate was 96.3% (52 of 54 cases). Six cases were excluded because biliary stents were placed after balloon dilation (Fig. 2).

Ultimately, the study included 46 patients for whom technical success was achieved. The baseline and perioperative characteristics are summarized in Table 1. Clinical success was achieved in all the cases. Recurrent HJAS was observed in 16 (34.8%) patients. The median follow-up period was 25.5 months (interquartile range, 15.2–41.1 months), and no significant difference in the follow-up time was observed between the two groups (26.9 months in the recurrent HJAS group vs. 24.0 months in the no recurrence group, p=0.76). Resected lesions were malignant in 75.0% of the recurrent HJAS group vs. 86.7% of the no recurrence group (p=0.42). Pancreaticoduodenectomy was the most common surgery, while hepatectomy was performed in 25.0% of the recurrent HJAS group vs. 10.0% of the no recurrence group (p=0.22). The most common suture method for hepaticojejunostomy anastomosis was a combination of continuous and discontinuous sutures; continuous sutures were used in only one case. No significant differences in preoperative bile duct diameter and preoperative biliary drainage were observed between the two groups. The median time from surgery to HJAS was shorter in the recurrent stenosis group than that in the no recurrence group (p=0.09).

Table 2 summarizes the characteristics of BAE-ERCP in the patients with and without recurrent stenosis. The presence or absence of a scar-like appearance at the anastomotic site was not associated with recurrent stenosis (56.3% with scar-like appearance vs. 36.7% without scar-like appearance, p=0.23). No significant differences in endoscopes used or use of a stent retriever were observed between the two groups. The length of the anastomotic stricture tended to be longer in the recurrent HJAS group than in the no recurrence group (4.5 mm vs. 2.5 mm, respectively; p=0.07). The median balloon diameter was 8.0 mm. Residual balloon notch was observed significantly more often in the recurrent stenosis group than in the no recurrence group (56.3% vs. 23.3%, p=0.049). Intrahepatic stone removal after balloon dilation of HJAS was performed in 43.8% of the recurrent HJAS group and 60.0% of the no recurrence group (p=0.36).

Figure 3 presents cumulative HJAS patency after balloon dilation. Patency rates after 1 and 2 years were 76.8% and 64.2%, respectively.

Table 3 presents the univariate and multivariate analyses of factors associated with recurrent stenosis after balloon dilation. Residual balloon notch was a significant predictor of recurrent HJAS (p=0.02), while HJAS within 1 year postoperatively (p=0.05) and anastomotic stricture length >3 mm (p=0.10) tended to predict recurrent HJAS in the univariate analysis. Meanwhile, the multivariate logistic regression analysis revealed that residual balloon notch was an independent factor for recurrent stenosis (HR, 2.80; 95% CI, 1.01–7.78; p=0.048). HJAS within postoperative 1 year tended to be associated with recurrent stenosis (HR, 2.43; 95% CI, 0.85–6.89; p=0.096). The length of the stenosis did not affect the restenosis rate. The patency time of patients with the disappearance of the balloon notch was significantly longer than that of patients with a residual balloon notch (p=0.02, Fig. 4). The patency rates of patients with and without a residual balloon notch were 66.7% and 82.1% after 1 year and 45.7% and 73.1% after 2 years, respectively.

Three AEs (6.5%) were observed: two cases of mild cholangitis and one case of perforation. No AEs were associated with scope insertion.

In the perforation cases, HJAS occurred 2.7 months after pancreaticoduodenectomy, and balloon dilation was performed. An 8-mm balloon was selected because the bile duct diameter was 8.8 mm, although a residual balloon notch was observed upon inflation. Three days after the procedure, the patient developed fever that was not accompanied by abdominal pain or jaundice. The patient’s C-reactive protein level was elevated, and computed tomography revealed a small amount of free air in the retroperitoneum. The patient was treated with antibiotics, and recovered with conservative treatment alone.