Endoscopic retrograde cholangiopancreatography (ERCP) is the gold standard for palliative treatment to relieve biliary obstructions in patient with unresectable malignancies.¹ However, it is sometimes very challenging to access or cannulate the major papilla for ERCP due to surgically altered anatomy or malignant duodenal obstruction. Under these circumstances, percutaneous transhepatic biliary drainage (PTBD) and enteroscopy (including single-balloon enteroscopy and double-balloon enteroscopy)-assisted ERCP (e-ERCP) have been widely performed as alternative procedures to ERCP.^2,3 While these techniques have several advantages, they also have disadvantages: PTBD is associated with a higher risk of complications and the need for external drainage, whereas e-ERCP is a time-consuming procedure and is associated with the risk of severe complications, such as perforation and pancreatitis.^4,5

In recent years, endoscopic ultrasound-guided hepaticogastrostomy with transmural stenting (EUS-HGS) has been reported as an alternative treatment for ERCP, PTBD, and e-ERCP.^6,7 EUS-HGS is feasible even in patients with a surgically altered anatomy or malignant duodenal obstruction. In addition, the risk of post-ERCP pancreatitis caused by traumatic papillary irritation was reduced in EUS-HGS.⁸ However, EUS-HGS is still challenging because of the complexity of the procedure, risk of severe adverse events, including stent migration, and lack of dedicated devices.⁹ In addition, There are several problems to be resolved before EUS-HGS can be optimized. One of these is the stent selection. Previous reports indicate that the majority of endoscopists prefer fully/partially covered self-expandable metallic stents (cSEMS) to plastic stents (PS)¹⁰ because they are considered to afford longer stent patency, reduced bile leakage, and easier reintervention.⁹ However, this preference is still based on the opinions of experts and evidence is still lacking, as no studies have been conducted to evaluate the advantages of cSEMS over PS for EUS-HGS in patients with unresectable malignancies.¹¹ Therefore, we performed a retrospective multicenter comparative study of cSEMS and PS to determine which of the two might be optimal for EUS-HGS in patients with unresectable malignancies.

EUS-HGS is often performed for palliation in patients with biliary obstruction due to unresectable malignancies, the majority of whom are undergoing chemotherapy. Long-term stent patency and low complication rates are crucial to maintain the performance status of these patients and to allow them to continue to receive chemotherapy and have the best chance of survival. In addition, as the prognosis of patients with unresectable malignancies has improved with advances in chemotherapy, the role of reintervention is also increasing. In view of the aforementioned issues, optimal stent selection is an issue that inevitably needs to be addressed. The use of cSEMS has gained popularity in patients undergoing EUS-HGS because it affords longer stent patency, reduced bile leakage, and easier reintervention.⁹ However, no comparative studies between PS and cSEMS have been performed; thus, evidence for the advantages of cSEMS is still lacking. This is the first study conducted in a relatively large number of patients from six participating facilities, to compare the TRBO, rate of complications, and rate of reintervention between patients receiving PS and cSEMS for EUS-HGS. In our cohort, multivariate analysis performed using ten variables identified two factors, including the use of cSEMS and EUS-HGAS, as independent predictors of favorable TRBO. We also demonstrated lack of significant differences in the rate of complications, including bile leakage and technical success rate of reintervention, between the PS and cSEMS groups.

Theoretically, EUS-HGS using a cSEMS is considered to afford longer stent patency due to the larger diameter of the stent, as in transpapillary biliary drainage.¹⁴ However, the differences between transpapillary biliary drainage and EUS-HGS still need to be considered, such as the location of the stent and whether the stent traverses a malignant stricture, which could affect the risk of RBO. Our results in this study support that even in the case of EUS-HGS, cSEMS affords longer stent patency than PS in patients with unresectable malignancies.¹¹ Recently, EUS-HGAS has been reported as an advanced palliative technique for obtaining long-term stent patency. Ogura et al.¹³ reported the clinical benefits of long stent patency in 49 patients who underwent EUS-HGAS, and mentioned that double-stent deployment could offer longer stent patency than EUS-HGS alone. Indeed, our study identified EUS-HGAS as an independent predictor of favorable TRBO.

No significant difference was observed in OS between the two groups. Therefore, the impact of patient death on TRBO is not biased. In addition, while the TRBO was significantly shorter in the PS group than in the cSEMS group, the patients’ survival in the latter part of the Kaplan-Meier curve was not significantly different. This suggests that the management of RBO in patients might be appropriate and crucial. The functional success rate in the cSEMS group was higher than that in the PS group (85% and 95%, respectively; p=0.130), although the difference was not statistically significant. This result could also be attributed to the larger diameter of the cSEMS.

In the present study, approximately one-third of the patients in both groups developed RBO, mainly due to sludge. However, the type of stent may also play a role, at least in part; RBO caused by tissue hyperplasia was significantly more frequent in the cSEMS group than in the PS group (7% and 0%, respectively; p=0.014). Nakai et al.¹⁵ reported that tissue hyperplasia is the major cause of RBO in patients undergoing EUS-HGS with cSEMS. This is generally believed to be caused by a mismatch between the size of the stent and the luminal diameter of the intrahepatic bile duct, although this hypothesis has not yet been validated and further study is required. Stent migration was observed in four patients, including two from the PS group and two from the cSEMS group. In all cases, the stent migrated to the gastrointestinal side. Interestingly, two patients in the cSEMS group who developed stent migration did not develop RBO, suggesting that the fistula might have been well sustained and served as a drainage route because of the large diameter of the cSEMS. In contrast, two patients in the PS group who developed stent migration developed RBO and required reintervention. In one of these patients, the fistula was identified endoscopically and an uncovered SEMS was deployed; in the other patient, the fistula could not be detected endoscopically, so additional EUS-HGS was performed with successful deployment of the PS. None of the patients developed RBO until the end of the follow-up period.

The reported rate of complications other than RBO after EUS-HGS in previous studies was 7% to 41%.¹⁶ From the standpoint of the risk of complications, especially bile leakage, cSEMS are considered preferable. However, endoscopists must be aware of the potential risk of cSEMS, including obstruction of the side branches of the biliary tract, which can lead to cholangitis or liver abscess. In our cohort, the rate of complications other than RBO was 16.4%, with no significant difference between the two groups. Consistent with previous reviews of EUS-HGS, the frequency of bile leakage was higher in the PS group than in the cSEMS group in this study, although the difference did not reach statistical significance.^9,10 Therefore, to prevent bile leakage, in addition to the selection of the appropriate stent, attention should also be paid to the distance from the intrahepatic bile duct to the liver surface. Yamamoto et al.¹⁷ reported that a short distance of the tract in the liver parenchyma (<2.5 cm) is an independent predictor of the risk of bile leakage. This is because the liver parenchyma may exert tamponade effects. We also encountered two patients who developed a liver abscess in the left hepatic lobe after cSEMS placement in B2. Considering the location of the abscess, we surmised that it might have been caused by blockage of side branches. A partially covered SEMS (pcSEMS) is often used to avoid blockage of the biliary side branches. However, it might still be challenging to prevent side branch blockage because the uncovered portion in a conventional pcSEMS is no greater than 5 mm in length. In EUS-HGS, the stent length in the intrahepatic bile duct should be at least 20 mm to prevent migration. Thus, even if a pcSEMS is used, the covered portion is deployed into the intrahepatic bile duct, which is associated with the risk of side-branch blockage. In fact, the pcSEMS was deployed in both patients. Recently, a dedicated pcSEMS (EGIS biliary stent, double-covered; S&G Biotech Inc., Seoul, Korea) for EUS-HGS was developed.¹⁸ Compared to the conventional pcSEMS, the uncovered portion at the distal end of this new stent is sufficiently long to prevent side branch obstruction. In addition, the stent had a radiopaque marker, indicating the end of the uncovered portion. These advantages may reduce the risk of side-branch blockage. Mediastinal emphysema was observed during the procedure in one patient in the cSEMS group. The transesophageal approach has been reported to carry the potential risk of severe complications, including mediastinal emphysema, mediastinitis, and pneumothorax.¹⁹ In this patient, B2 was punctured via the gastric wall just below the esophagus, and the tract was subsequently dilated with an electrocautery dilator. When puncturing B2, we carefully visualized the crus of the diaphragm using EUS to avoid mediastinal punctures. However, both the damaging effect of the self-expandable system of cSEMS and the burning effect of the electrocautery dilator on the diaphragm and the consequent risk of mediastinal emphysema should also be considered.

In the present study, no significant difference was observed in the technical success rate of reintervention or the time taken for the reintervention procedure between the two groups. However, it is worth noting that the technical success rate of the reintervention was 100% in the cSEMS group. In patients in whom a cSEMS was used, reintervention, including stent-in-stent placement and balloon sweeping, was performed using the existing cSEMS. Technically, it was relatively easy to perform because of the large diameter of the cSEMS. In contrast, in patients in whom PS had been used, reintervention usually necessitated stent exchange. From this viewpoint, planned stent exchange of the PS with the cSEMS could contribute to reducing the incidence of cholangitis secondary to stent occlusion. Before the PS is changed, a guidewire needs to be introduced into the intrahepatic bile duct through or along the PS using an ERCP catheter. If this fails, the PS should be retrieved, followed by detection of the fistula using the guidewire. These procedures are technically challenging and could have contributed to the lower success rate of reintervention in the PS group. As the number of patients who underwent reintervention was small in this study, further accumulation of patients would be desirable to obtain more robust results.

This study has several limitations. First, it was a non-randomized retrospective study, which is often associated with a selection bias. Thus, various types of cSEMS and PS were used in this study. A randomized controlled study is required to draw definitive conclusions about the optimal stent selection for EUS-HGS for malignant biliary obstructions. However, we included a relatively large number of patients (n=154) in this study, and EUS-HGS was performed by different endoscopists at the six participating facilities, which could contribute to the validity of our results and show real-world results. Second, the EUS-HGS procedure used, including the bile duct segment puncture and the tract dilator device used, differed significantly between the two groups, which could have influenced the outcomes of EUS-HGS. In terms of TRBO, our multivariate analysis revealed that neither factor had a significant effect on this outcome. However, we could not analyze the influence of these two factors on the rate of complications. A previous study reported that the use of an electrocautery dilator device, including a needle knife, was associated with a higher rate of complications.^20-22 Therefore, a high-quality study in a larger cohort is required using a design that excludes the effect of the tract dilator used on the risk of complications. Third, due to the nature of unresectable malignancies, 50 patients (37.3%) died (censored) without RBO. This resulted in an increase in variability in the latter part of the Kaplan-Meier curves of TRBO.

Our study suggests the advantages of cSEMS over PS for EUS-HGS in patients with unresectable malignancies regarding RBO, as shown in similar previous studies. Therefore, it is anticipated that newer cSEMS specifically designed for EUS-HGS can be used to standardize the use of cSEMS.