The increased use of neoadjuvant chemotherapy for biliary and pancreatic cancers and longer preoperative waiting periods have increased the need for preoperative biliary drainage (PBD) for patients with distal or proximal biliary obstruction secondary to a resectable tumor.^1,2 Although the endoscopic transpapillary approach of endoscopic retrograde cholangiopancreatography (ERCP) is the established technique for PBD, endoscopic ultrasound (EUS)-guided biliary drainage (EUS-BD) has recently emerged as a useful alternative after failed ERCP.³ Percutaneous transhepatic biliary drainage is the classic alternative to PBD with ERCP, but this technique is limited by difficulties associated with single-procedure internal fistula formation and outpatient self-management of cases of external fistula formation. Fortunately, EUS-BD may address these problems.

EUS-BD comprises various options, including hepaticogastrostomy (HGS), choledochoduodenostomy (CDS), and antegrade stent placement (AGS), as well as the use of plastic and metallic stents; however, an optimal method for PBD has not been identified. In addition, the efficacy and safety of EUS-BD and its impact on surgery remain unclear. This study presents a case in which EUS-HGS was effective and safe for salvage drainage of a resectable pancreatic head tumor that was difficult to cannulate using transpapillary biliary drainage during ERCP.

A 76-year-old man who had undergone endoscopic biliary stenting at a previous hospital because of obstructive jaundice attributable to distal bile duct stenosis caused by carcinoma of the pancreatic head was transferred to our hospital because of complications during stent replacement. Cholangitis caused by stent obstruction occurred immediately after the start of neoadjuvant chemotherapy. Stent replacement was attempted, but the bile duct orifice was difficult to visualize endoscopically because of duodenal stenosis caused by tumor invasion; therefore, the stent could be removed but not reinserted. The patient had undergone distal gastrectomy (Billroth I reconstruction) for advanced gastric cancer 4 years previously.

At the time of transfer, the patient was in stable condition without abdominal pain. His blood pressure was 154/81 mmHg, heart rate was 69 beats/min, and temperature was 36.7°C, probably because he was fasted and received rehydration fluids and antibiotics at the previous hospital. His serum total bilirubin and C-reactive protein levels were 9.2 mg/dL and 11.99 mg/dL, respectively.

Triphasic computed tomography of the abdomen revealed a poorly enhanced tumor in the pancreatic head, overlapping stenosis of the distal bile duct and duodenum, and dilatation of the upstream common bile, bilateral intrahepatic bile, and caudal pancreatic ducts (Fig. 1).

Figure 1. Triphasic computed tomography image after hospital transfer. (A) A poorly enhanced tumor is detected in the pancreatic head (yellow arrowheads) with an increased fatty tissue concentration around the tumor. (B) The caudal side of the tumor shows atrophy of the pancreatic parenchyma. (C) The tumor invades the distal bile duct and descending part of the duodenum. (D–F) Intrahepatic bile duct dilatation caused by bile drainage obstruction attributable to the tumor in the coronal section.

Percutaneous transhepatic biliary drainage and EUS-HGS were proposed as salvage biliary drainage options; however, EUS-HGS was selected because the patient strongly preferred single-procedure internal fistulation so he could continue neoadjuvant chemotherapy on an outpatient basis. A self-expandable metallic stent (SEMS) (Niti-S EUS-BD system; Taewoong Medical) was placed from the residual stomach to the left intrahepatic bile duct (B3) (Fig. 2). This stent had an 8-mm lumen with a 15-mm uncovered portion on the hepatic side and an anti-migration system (spring stopper) on the gastric side. After considering that the stent may need to be removed during surgery, the uncovered part of the stent on the bile duct side was placed slightly more toward the B3 bile duct side than toward the B2-B3 bile duct bifurcation so it would not bend. The computed tomography findings observed the day after EUS-HGS indicated no signs of inflammation or fluid collection in the fistula formation area. The patient recovered without complications, resumed eating on day 2, and was discharged on day 8 after the procedure. At the time of discharge, the serum total bilirubin and C-reactive protein levels had improved to 2.1 mg/dL and 1.8 mg/dL, respectively. The patient completed a total of approximately 2 months of neoadjuvant chemotherapy; during that time, stent obstruction, cholangitis, and other complications did not occur.

Figure 2. Self-expandable metallic stent placed from the residual stomach to the left intrahepatic bile duct. (A) Representative fluoroscopic image. (B) Coronal section of the computed tomography image.

Robotic-assisted laparoscopic pancreaticoduodenectomy was performed 39 days after EUS-HGS (Fig. 3). Extensive adhesions from the subhepatic space to the dorsal surface of the lateral zone were present, and a clean fistula had formed at the metallic stent. The adhesions were observed not only on the lower surface of the liver but also throughout the upper abdomen; they were considered attributable to gastrectomy that had been performed 4 years previously. The stent was exposed by laparoscopic dissection and removed from the gastric and bile duct sides, in that order. Although the anti-migration system provided some resistance, stent removal was easily performed by grasping it with forceps and pulling it vertically against the stomach wall. Therefore, surgery was successfully performed without damaging the organs or vessels on either side. Subsequently, the fistula was sutured to prevent postoperative fluid leakage. The time required to remove the SEMS was 5 minutes and 35 seconds (from the start of dissection around the fistula to fistula exposure: 1 minute and 10 seconds; from fistula exposure to stent removal: 4 minutes and 25 seconds). The time required to close the fistula was 8 minutes and 40 seconds (hepatic side: 5 minutes and 55 seconds; gastric side: 2 minutes and 45 seconds). No indications of tumor dissemination were intraoperatively observed. The patient had a good postoperative course without complications such as bile leakage and peritonitis, and he was discharged on postoperative day 16. A postoperative pathological examination revealed well-differentiated to moderately differentiated invasive ductal carcinoma.

Figure 3. Intraoperative findings of robot-assisted laparoscopic pancreaticoduodenectomy. (A) The hepaticogastrostomy fistula was identified after detaching the adhesions from the subhepatic space to the dorsal surface of the lateral hepatic zone. (B) The fistula was dissected to expose the self-expandable metallic stent. (C) Some resistance occurred when removing the stent on the gastric side. (D) An anti-migration system in the stent removed on the gastric side. (E) The stent on the bile duct side was relatively easily removed. (F) The fistula was sutured after stent removal to prevent postoperative fluid leakage.

The participant provided informed consent.

Recently, EUS-BD has been applied as a salvage method after failed ERCP. The types of EUS-BD that are considered most effective for PBD are HGS, CDS, AGS, and gallbladder drainage.^4–7 For our case, HGS was selected because tumor invasion created difficulty when manipulating the endoscope in the duodenum and it has a limited influence on the surrounding pancreatic head during surgery. EUS-HGS forms a fistula between the stomach and left lobe of the liver and may have less of an anatomical impact on the operative area during surgery than that of CDS and AGS for cases that require pancreaticoduodenectomy, such as pancreatic or distal bile duct cancer. Mukai et al⁴ retrospectively evaluated the clinical feasibility of preoperative EUS-BD, mainly EUS-HGS, for PBD in patients with malignant biliary obstruction and reported successful stenting within a short time and clinical improvement (lower bilirubin levels and cholangitis control) without stent-related adverse events, recurrent biliary obstruction, and surgery-related adverse events.

Several studies have confirmed the efficacy and safety of CDS and AGS for PBD. For example, a study that compared the efficacy and safety of PBD using ERCP and those of PBD using CDS with an electrocautery-enhanced lumen-apposing metallic stent in patients before pancreaticoduodenectomy found that CDS deployment was easier and more efficient than ERCP; additionally, CDS resulted in fewer surgical complications without compromising the oncologic outcome.⁸ EUS-AGS effectively improved the technical success, clinical success, EUS-BD procedure-related complications, surgery-related complications, and 3-month postoperative mortality of patients receiving preoperative and palliative care.⁵ Therefore, although CDS and AGS are safe and effective for PBD, unlike HGS, they may not be indicated when intestinal stenosis is present in the duodenum or main papillary region, as in our case. Selecting the appropriate drainage method according to the clinical condition of the patient is important.

Endoscopic stent removal using EUS-HGS after surgery is a viable option. However, when using endoscopy to remove the stent, it is not possible to detect or close the fistula on the hepatic side after removal. Therefore, after considering the risk of bile leakage, we decided that intraoperative removal was the superior method and used it for this case. During preoperative EUS-HGS for distal malignant biliary obstruction, the stent placement site is not included in the resection range; therefore, it is necessary to avoid postoperative complications and minimize the impact of surgery on the primary lesion. Unlike CDS and AGS, surgery after EUS-HGS requires identification of the fistula site and removal of the stent in addition to resection and reconstruction of the main lesion. Therefore, the stent used for EUS-HGS may be the one that is easy to remove (such as a plastic stent or fully covered metallic stent). However, when EUS-HGS is performed for operable patients, it is essential to minimize the risk of complications such as bile leakage and peritonitis to allow continuation of chemotherapy and prevent dissemination. The risks of bile leakage and stent migration may be higher with plastic stents and fully covered metallic stents; therefore, for this case, we chose a SEMS with a partially uncovered bile duct side and an anti-migration system on the gastric side. These details regarding stent selection must be fully explained to and understood by the patient before surgery, and informed consent must be obtained. Although we performed robot-assisted laparoscopic surgery, which has a limited field of vision compared to that of open surgery, and although we used a SEMS with an anti-migration design, the fistula in our patient was quickly identified and the stent was safely removed. Furthermore, peritonitis, bile leakage, and dissemination attributable to EUS-HGS did not occur. Similarly, the fistula was intraoperatively identified and the stent was safely removed in another patient with obstructive jaundice caused by resectable bile duct cancer and treated with PBD using EUS-HGS.⁹ Adverse events related to stent removal for cases treated with preoperative EUS-HGS (cases in which the stent could not be safely removed, cases with bile leakage development, and others) have not been reported, possibly because of significant publication bias; therefore, larger prospective studies of preoperative EUS-HGS are necessary.

In conclusion, we effectively and safely applied EUS-HGS for PBD of a malignant biliary stricture. Although the efficacy and safety of EUS-BD for PBD have been widely report, most studies have been small and retrospective, and the safety of intraoperative stent removal using HGS remains unclear. This is the first report of the safety of intraoperative stent removal using EUS-HGS for PBD. Future case series and large prospective studies are needed.

None.

The data are available from the corresponding author upon reasonable request.

No potential conflict of interest relevant to this article was reported.