Malignant biliary obstruction (MBO) is a common but still challenging condition often encountered in cases with pancreatobiliary malignancies. Endoscopic retrograde cholangiopancreatography (ERCP) has long been the mainstay for biliary drainage due to its less invasiveness. However, ERCP can fail in about 5 to 10%, both due to technical difficulty and anatomical reasons such as duodenal obstruction and altered anatomy. Percutaneous transhepatic biliary drainage (PTBD) is often utilized in cases after failed ERCP but endoscopic ultrasound-guided biliary drainage (EUS-BD) has emerged as an alternative to PTBD. EUS-BD was first reported as a rescue technique but can be increasingly utilized as a primary technique as the new dedicated devices develop and the techniques improve. There are some drainage approaches such as EUS-guided choledochoduodenostomy (EUS-CDS) and hepaticogastrostomy (HGS) as well as clinical situations such as preoperative and palliative biliary drainage. This review aims to discuss current evidence, hurdles and future perspectives of primary EUS-BD for MBO.

EUS-BD was first introduced as a rescue technique after failed ERCP [1], while the conventional alternative procedure after failed ERCP used to be PTBD. In a meta-analysis of EUS-BD and PTBD after failed ERCP [2], there were no significant differences in technical success with an odds ratio (OR) of 1.78 (95% confidence interval [CI], 0.69-4.59). When limited to 3 randomized controlled trials (RCTs), OR was 0.68 (95% CI, 0.14-3.27), favoring EUS-BD without statistical significance. Furthermore, clinical success was significantly better in EUS-BD with an OR of 0.45 (95% CI, 0.23-0.89) and the adverse event (AE) rates were also lower in EUS-BD with an OR of 0.23 (95% CI, 0.12-0.47).

One of the disadvantages of PTBD is the discomfort by tube placement, affecting the quality of life. In an international multicenter survey for preference between EUS-BD and PTBD [3], 80.2% of patients preferred EUS-BD rather than PTBD. Patients preferred EUS-BD because of less discomfort, higher success with less morbidity, and one-stage internal drainage procedure. Meanwhile, PTBD was preferred because of proven technical safety, short procedure time, technical easiness and less cost.

Thus, safety and technical difficulty of EUS-BD, especially in the absence of expertise is still the concern. Among three meta-analyses of EUS-BD, AE rates have decreased chronologically; 23.32%, 17.9%, and 13.7% reported in 2016 [4], 2020 [5], and 2023 [6], respectively. The rates of bile leak also decreased; 4.03%, 4.1% and 2.2%, respectively. The development of various devices as well as the established techniques may contribute to this better safety for EUS-BD. Given its improved safety, EUS-BD is now considered as a rescue to failed ERCP when expertise is available.

ERCP is the mainstay of biliary drainage for MBO, but transpapillary approach by ERCP-BD has its inherent disadvantages; Failed biliary cannulation and post-ERCP pancreatitis (PEP). Biliary cannulation is not always successful even by experts. In cases with MBO, technical difficulty can increase due to the distorted anatomy by cancer invasion. In cases with duodenal invasion by pancreatic cancer, duodenobiliary reflux can cause cholangitis or stent occlusion by food debris, even if ERCP is technically possible. In a retrospective study of ERCP-BD using metal stents [7], duodenal invasion was a risk factor for early stent dysfunction with an OR of 2.35. EUS-BD, especially EUS-HGS, can avoid duodenobiliary reflux and provide longer time to recurrent biliary obstruction (RBO). EUS-HGS showed less early RBO compared to ERCP-BD in cases with duodenal invasion [8]. Surgically altered anatomy is also a technical hurdle to achieve biliary access. In cases with surgically altered anatomy, balloon endoscope-assisted ERCP is performed but scope insertion or biliary cannulation can be technically difficult, depending on the prior surgical procedures. Technical success rate was low in cases with gastric bypass with Roux-en-Y reconstruction due to its long limb [9], or with an intact papilla due to difficult cannulation [10]. Thus, EUS-BD for those cases provides higher technical success rates.

PEP is also an unsolved problem with ERCP, especially in cases without cancer involvement to the main pancreatic duct [11]. As stents do not cross the papilla in EUS-BD, there are almost no risks of post-procedure pancreatitis. Furthermore, stent placement across the stricture in ERCP-BD increases the risk of stent occlusion by tumor ingrowth and overgrowth. On the other hand, tumor-related stent occlusion is rare in EUS-BD and EUS-BD can theoretically provide longer time to RBO than ERCP-BD.

According to the advantages and disadvantages of each procedure, preferred clinical scenarios for EUS-BD and ERCP-BD are summarized in Table 1.

Given the potentially longer stent patency and improving safety, EUS-BD as a primary drainage technique for MBO has been reported in many clinical trials. Recently, a meta-analysis of RCTs comparing EUS-BD vs. ERCP-BD for MBO [12] was reported. Risk ratios (RRs) for technical and clinical successes were 1.06 (95% CI, 0.96-1.17) and 1.02 (95% CI, 0.97-1.08). As for safety, RRs for overall and severe AEs were 0.85 (95% CI, 0.49-1.46) and 0.97 (95% CI, 0.10-9.17), suggesting comparable short-term outcomes. As for long-term outcomes, 1-year stent patency was similar with RR of 1.15 (95% CI, 0.94-1.42) but reintervention rates were lower in EUS-BD (RR, 0.58; 95% CI, 0.37-0.9).

In the subgroup analysis, technical success rate was higher in EUS-BD with Lumen-apposing metal stent (LAMS) than in ERCP-BD, with RR of 1.17 (95% CI, 1.01-1.35). Meanwhile, LAMS was not superior in terms of long-term outcomes. RRs for 1-year stent patency were 1.48 (95% CI, 0.97-2.24) for EUS-BD with self-expandable metallic stent (SEMS) and 1.02 (95% CI, 0.94-1.10) for EUS-BD with LAMS. Reintervention rates were also lower only in EUS-CDS with SEMS (RR, 0.40; 95% CI, 0.23-0.71) but not in EUS-CDS with LAMS group (RR, 0.91; 95% CI, 0.48-1.73), suggesting disadvantages of LAMS in terms of long-term outcomes.

Finally, a dilated (>15 mm) common bile duct was associated with technical success of EUS-BD with LAMS. When LAMS is used for EUS-CDS, sufficient bile duct dilatation is mandatory to deploy its large flange. However, in patients with distal MBO, the bile duct diameter was less than 12 mm in 44.9% [13], suggesting EUS-CDS with LAMS is not always technically feasible. Recent two studies [14,15] on stent misdeployment of EUS-CDS with LAMS showed that the bile duct diameter <15 mm was a risk factor for misdeployment or technical failure. Although most of those complications can be managed endoscopically, some cases need percutaneous or surgical procedures.

In summary, while EUS-CDS with LAMS can provide better short-term clinical outcomes, it is only technically feasible in cases with a markedly dilated bile duct and its long-term outcomes are not necessarily superior to EUS-CDS with tubular SEMS.

There are various approaches for EUS-BD. EUS-CDS and EUS-HGS are two major procedures for EUS-BD but recently EUS-GBD is also reported as a rescue when EUS-CDS or EUS-HGS is difficult. In a meta-analysis of 225 EUS-HGS and 312 EUS-CDS [16], there were no significant differences in the technical success (OR, 0.83; 95% CI, 0.41-1.68), clinical success (OR, 0.96; 95%CI, 0.51-1.81) but AEs were more common in EUS-HGS (OR, 2.01; 95% CI, 1.14-3.59). Stent patency is also comparable between EUS-CDS and EUS-HGS [17]. Thus, if both EUS-CDS and HGS are technically feasible, EUS-CDS can be a first-line technique for primary EUS-BD. However, as mentioned above, in cases with duodenal invasion or surgically altered anatomy, EUS-HGS is the treatment of choice.

EUS-CDS and EUS-HGS can be technically difficult due to the small bile duct and intervening vessels or cystic duct. The usefulness of EUS-GBD as a rescue is increasingly reported. In a recent meta-analysis, technical and clinical success rates were 99.2% and 88.1%, respectively, and AE rate was 13.7% [18], which is comparable to EUS-CDS and EUS-HGS. In a propensity score-matched cohort study of EUS-GBD and EUS-CDS [19], technical and clinical success rates were comparable: 96% vs. 99% for technical success and 86% vs. 92% for clinical success, respectively. AE rates were 14% both in EUS-GBD and EUS-CDS. Thus, EUS-GBD can also be a treatment option as primary biliary drainage for distal MBO, though cystic duct patency needs to be confirmed prior to EUS-GBD [20]. Given the large size of gallbladder and the availability of LAMS, EUS-GBD might be technically less demanding.

Despite the increasing evidences of EUS-BD, ERCP-BD is still the mainstay of biliary drainage for MBO. In an international survey of 115 gastroenterologists and surgeons conducted in 2019 [21], 81.7% considered EUS-BD to have either equivalent or better technical and clinical success than ERCP. However, only 11% of respondents would consider EUS-BD as a first-line modality regularly, and 35% never considered EUS as a first line modality. Major concerns against the use of EUS-BD were the lack of high-quality data (40.9%), fear of AEs (36.5%), and limited access to EUS or dedicated devices (29.6%). Since the increasing evidences of safety and efficacy of EUS-BD since 2019, the current concern would be the limited access to EUS and dedicated devices. In EUS-CDS as a primary technique, the use of LAMS was associated with higher technical success and shorter procedure time [12].

The remaining major issue for wide spread adoption of EUS-BD lies in the access to EUS-BD procedures as well as technical difficulty for trainees. A few studies evaluated the learning curves of EUS-BD. In one study, technical success rate of EUS-HGS was 100% after 40 cases [22]. In another study, procedure time was shorter and the AE rates were lower after 33 cases of EUS-HGS [23]. However, the number of EUS-BD procedures is not large enough to achieve this kind of learning curves, except for high-volume centers. A recent Japanese study evaluating clinical outcomes of the initial 20 EUS-BD procedures demonstrated a high technical success rate of 91.4% with an AE rate of 10.2% [24]. Another Japanese study [25], suggested the experiences in each institution, in addition to personal experiences, can affect clinical outcomes of EUS-BD. In addition to the personal experiences of 436 screening EUS, 93 EUS-guided fine needle aspiration (EUS-FNA), and 13 EUS-guided drainage, institutional experiences of 7 or more EUS-BD were associated with successful EUS-BD.

If all ERCP-BD were to be replaced by primary EUS-BD, those numbers of 30-40 cases can be performed to achieve the learning curves by trainees. Otherwise, it is difficult for a single trainee to experience such a number of procedures, except for high volume centers. To overcome this issue, training models for EUS-BD have been developed and utilized in the hands-on course [26,27]. While the usefulness of training models was reported [28], the opportunity for performance of EUS-BD was limited. Ten out of 17 trainees did not perform EUS-HGS after hands-on program, and even among 7 trainees who did perform EUS-HGS, the first procedure was later than six months after hands-on program in 2 trainees. Given the limited number of EUS-BD procedures, the appropriate timing and applicants for the hands-on course are yet to be established. Furthermore, credentialing and competency assessment need to be standardized.

Despite wide spread use of EUS-BD procedures, there is still some controversy on primary EUS-BD for MBO. First, development of dedicated devices is essential for safe and effective procedures. Currently, most devices are originally developed for EUS-FNA or ERCP, not dedicated for EUS-BD. Recently, however, dedicated stents were available such as LAMS [29] or plastic stent [30]. In addition, various dilation devices [31-34] have been developed but are only available in limited countries.

Second, the feasibility of preoperative EUS-BD as an expanded indication is currently under debate. AEs of biliary drainage procedures can affect subsequent surgical procedures such as increased blood loss and operative time. In cases with potentially resectable pancreatic cancer, metal stents are increasingly placed to cover the neoadjuvant treatment period, but are associated with PEP [35]. Since EUS-BD does not traverse the ampulla, there is no risk of post-procedure pancreatitis. In a recent propensity score-matched study [36], EUS-CDS did not interfere with pancreatoduodenectomy without an increase in postoperative complications. Furthermore, transpapillary stent placement per se can cause inflammation around the bile duct and sometimes make surgical procedures difficult. EUS-HGS, which drains the bile duct away from the surgical site, can avoid inflammation around the head of pancreas and is also reported to reduce bacterial contamination in the bile duct [37]. However, there are potential disadvantages in EUS-BD. Since the procedure includes the puncture of the bile duct, there is always a risk of bile leakage, which can cause inflammation or even tumor seeding. Since PTBD for bile duct cancer was oncologically inferior to ERCP-BD [38], long-term outcomes of surgical resection after EUS-BD should be further investigated.

Finally, cost-effectiveness analyses are important since EUS-BD is in general an expensive procedure, especially when LAMS is used. If EUS-BD can provide longer time to RBO, cost of EUS-BD procedures with devices can be mitigated by less reinterventions and reduced hospitalization. Reimbursement is another issue for EUS-BD, which can be a hurdle for physicians to perform this procedure.

In summary, EUS-BD has gradually transitioned from rescue procedures after failed ERCP to primary drainage procedures. The major advantages of EUS-BD are internal biliary drainage both without traversing the biliary stricture and the papilla. In cases with distal MBO, EUS-CDS with LAMS can be a primary technique if the bile duct is dilated enough. Meanwhile, in cases with duodenal invasion or surgically altered anatomy, EUS-HGS is preferred. Although clinical trials demonstrated EUS-BD can provide better clinical outcomes than ERCP-BD, this advantage is observed only when expertise is available. Thus, training and credentialing for EUS-BD are mandatory to further expand generalizability of this promising procedure.