The procedural steps of EUS-HG are as follows: Using an interventional echoendoscope, the dilated left hepatic duct (usually segment III) is well visualized. EUS-HG is then performed under combined fluoroscopic and ultrasound guidance, with the tip of the echoendoscope positioned such that the ultrasound transducer is either in the middle part of the small curvature of the stomach or slightly upwards, closer to the cardia. A 19G FNA needle is inserted transgastrically into a peripheral branch of the left hepatic duct, and contrast medium is injected. Before injecting the contrast fluid, bile can be aspirated through the needle in order to confirm the intraductal position of the needle tip. Opacification delineates fluoroscopically the dilated biliary tree above the point of obstruction. The needle is exchanged over a guidewire for a 6.0-Fr diathermic sheath, which is then used to enlarge the channel between the stomach (or jejunum in patients with total gastrectomy), and the left hepatic duct. The diathermic sheath is advanced across the intervening liver parenchyma by using a cutting current. After removing the guidewire, the diathermic sheath, which is an 8.5-Fr, 8-cm - long hepatico-gastric stent) or an 8 cm long covered self-expandable metal stent (SEMS), partially covered or fully covered, is placed transmurally. Fluoroscopy confirms adequate stent placement and function by showing contrast drainage through the stent into the stomach (Fig. 1).

Bile leakage into the peritoneum is the major risk factor for EUS-HG. Several strategies are used by different authors to minimize this risk. A 6 or 7-Fr naso-biliary drain with mild aspiration or gravity drainage can be left in place through the metal stent over 48 hours, even if this is somewhat inconvenient to the patient. More recently, this method has been developed into a more patient-friendly approach to minimize the risk of leakage, by combining an uncovered metal stent with a covered metal stent inside. The uncovered stent is initially deployed to provide anchorage and prevent migration. The covered stent is inserted coaxially and deployed within the first stent. Finally, in cases where the guidewire crosses the downstream stricture antegradely, hepaticogastrostomy can be combined with antegrade placement of an additional metal stent bridging in the distal stricture, which further decreases the pressure gradient across the transmural stent by providing additional downstream decompression of the bile duct (6). Alternative strategies used by other authors to prevent migration include the placement of fully covered SEMS with both ends flared (7), or forceful balloon expansion upon stent deployment (as opposed to gradual spontaneous self-expansion over several hours) to monitor foreshortening and the insertion of a double pig-tail stent through the expanded SEMS in order to provide additional anchorage (8).

The specific anatomic features of patients that may make EUS-HG preferable to other EUSBD are based on the intrahepatic access route and the transmural drainage route. Intrahepatic access is the only choice in patients with proximal (hilar) biliary obstruction, and is usually more convenient in patients with distal gastrectomy since imaging the common bile duct (CBD) under EUS is not always possible in postoperative altered anatomy setting (9). One advantage of transmural drainage after intrahepatic bile duct access over transpapillary drainage is that the challenging step of antegrade guidewire passage (required for both rendezvous and antegrade stenting) is avoided. In addition to guidewire passage, rendezvous requires an accessible papilla, which is usually not the case in patients with a surgically altered anatomy or tight duodenal stenosis. Antegrade stent insertion does not require an accessible papilla, but involves dilation of the puncture tract, just as EUS-HG. In patients with postoperative anatomy, antegrade transpapillary stenting without combined hepaticogastrostomy is less convenient for stent revisions, since HG provides easy repeat access to the bile duct without the need for a repeat puncture. Stent revisions are not uncommonly required during follow-up. The advantages of EUS-HG over rendezvous or antegrade stent insertion are particularly relevant in patients with prior duodenal or biliary SEMS who experience recurrent biliary obstruction (10).

However, these variant EUSBD approaches should be viewed as complementary rather than mutually exclusive. For example, as mentioned when discussing strategies to minimize the risk of bile leakage in EUS-HG, antegrade transpapillary stents can be combined with transmural stenting (6). Püspök et al. (11) performed an antegrade transpapillary SEMS insertion in a patient with recurrent gastric cancer after a Roux-en-Y gastrectomy. The authors then left a transmural plastic stent across the puncture tract to both minimize the risk of leakage and to preserve access. Dual drainage (antegrade and transmural) has also been used serially. Fujita et al. (12) performed transesophageal EUSBD by inserting a 7-Fr plastic stent into a peripheral left bile duct branch in a patient with advanced gastric cancer. Ten days later, the plastic stent was cannulated with a guidewire and removed with a snare (13). Then, using flexible devices through the mature fistula, the guidewire was manipulated under fluoroscopy across the malignant distal bile duct stricture, and a SEMS passed antegradely over the wire was subsequently deployed across the stricture above the papilla.

Patients with distal bile duct obstruction without prior gastrectomy who have both intra and extrahepatic bile duct dilations (and no gross ascites), are the only cases in which an intrahepatic or extrahepatic access site for EUSBD might be preferable if the selection criteria for EUSBD versus PTBD are broad, [i.e., EUSBD is favoured as the initial second-line approach after failed ERCP, and this type of patient may represent just 20% of the candidate population (14)]. Operator preference plays a part in this small patient subset. CBD offers a more obvious target for an EUS puncture, the echoendoscope is in a more anchored position, and access to the CBD probably makes rendezvous easier than for intrahepatic access. On the other hand, intrahepatic EUSBD is performed with an echoendoscope in a more straight position, which favors the transmission of the pushing force during stent insertion. It is also probably easier to penetrate a small intrahepatic bile duct surrounded by liver parenchyma than the fibrotic, hard wall of the CBD.

To date, transmural intrahepatic EUSBD has been reported in 51 patients, EUS-HG in 42, and other closely related variant approaches through a transjejunal or a transesophageal route in 9. In five patients with total gastrectomy, the left bile duct was similarly accessed under EUS from below the cardia and transmural stents were placed across the jejunal wall. In the remaining four patients, a cephalad peripheral left bile duct branch was selected for puncture, so that eventually the stent pierced the wall of the intra-abdominal esophagus slightly above the cardia. Approximately half of these patients come from three small series, which specifically deal with transmural intrahepatic EUSBD (6, 10, 15), whereas the other half comes from either mixed series in which EUS-HG is reported along extrahepatic EUSBD (7, 11, 16-19) or individual case reports (12, 20-24) (Table 1).

EUS-HG (or its variants) was technically successful in 49 out of these 51 patients, with clinical resolution of biliary obstruction in 46 cases. Therefore EUS-HG had a 94% per-protocol success rate and a 90.2% success rate on an intention-to-treat basis. These success rates are very high, considering the difficult patient population in which EUS-HG was attempted. However, it is important to remember that success was achieved at the expense of an overall 20% complication rate, twice as high as that of ERCP. Most complications were accounted for by inadequate biliary drainage, resulting in either peritoneal bile leakage or cholangitis (Table 1). Plastic stents caused cholangitis due to early migration (19) or early clogging (6). Foreshortening of transmural SEMS led to bile perotinitis or biloma, requiring percutaneous drainage and repeat EUSBD (6), and caused the only reported death to date (24). Half of the complications were nonetheless mild, manifesting as transient abdominal pain with or without pneumoperitoneum that settled on conservative measures.

There is great consistency across all reports on EUS-HG regarding technical details except for the use of cautery, be it kneedle-knives or fistulotomes and the choice of the stent. Overall, any diathermy use was reported in just 39.5% of cases, with some authors using it routinely (6), and others resorting to it selectively (7) (only after failure to advance a mechanical dilator over the guidewire), or do not use it at all (15).

As detailed in Table 1, 7 to 8.5-Fr plastic stents were placed in 46% of cases, whereas uncovered, partially covered or fully covered SEMS were placed initially in 54% of cases. It is difficult to draw significant conclusions from published reports, since no formal comparisons have been made between the two types of stents. SEMS are appealing for three reasons. First, upon full expansion, SEMS effectively seals the puncture/dilation tract, which would in theory prevent leakage more effectively. Secondly, their larger diameter provides better long-term patency, which would decrease the need for stent revisions. Finally, if dysfunction by ingrowth or clogging occurs, management is somewhat less challenging than with plastic stents, since a new stent (plastic or SEMS) can easily be inserted through the occluded SEMS in place. In contrast, exchanging a clogged plastic transmural stent usually requires over-the-wire replacement, because free-hand removal involves the risk of track disruption with subsequent guidewire passage into the peritoneum, hence requiring repeat EUSBD (or PTBD) to re-establish drainage (13). These presumed advantages of SEMS must be balanced against the fact that transmural SEMS insertion and deployment are somewhat more demanding than they are for ERCP. In particular, the serious risk of foreshortening and bile peritonitis should be prevented with careful attention to detail (24).

EUS-guided extrahepatic bile duct drainage is performed in a procedure where the CBD is punctured from a transduodenal or transgastric approach (usually from the distal antrum). The CBD is more easily imaged under EUS than the intrahepatic bile ducts, providing a faster, cleaner access without repeated puncture attempts, thereby minimizing risks. The retroperitoneal location of the CBD makes it also an attractive access site for patients with ascites, in whom fluid around the liver makes transhepatic access (whether percutaneous or transgastric under EUS) more difficult and hazardous.

Besides the advantages of extrahepatic access over intrahepatic access, the specific rationale for EUS-Choledochoduodenostomy (CDS) is also derived from the transmural drainage route, as opposed to transpapillary EUSBD (antegrade or rendezvous) (11, 25). The real choice between transmural and transpapillary drainage after extrahepatic bile duct access under EUS, therefore lies between EUS-CDS and rendezvous. Proponents of rendezvous argue that it may be less invasive than EUS-CDS, since transmural interventions are usually limited to puncture and guidewire passage. Hence, drainage is accomplished retrogradely via ERCP without the need for puncture tract dilation (26). However, EUSBD rendezvous carries a 20% failure rate; even in expert centers, because guidewire passage across the stricture and the papilla is often unsuccessful. The needle allows virtually no interplay with the guidewire, which cannot be manipulated across the stricture through a needle in the same way as it can be done at ERCP using flexible catheters. Therefore, advantages of EUS-CDS over transpapillary rendezvous are its higher success rate and relative simplicity, which appear to make it a more reproducible approach, despite being perhaps more invasive. Nonetheless, both EUSBD variant approaches can be considered complementary inasmuch as these procedures are used in a heterogeneous patient population. As we will discuss below, some indications are better suited for EUS-CDS, whereas in other cases EUSBD rendezvous is clearly advantageous. Similarly, even if rendezvous is the intended drainage technique, EUS-CDS can be used as a second line approach to salvage the significant proportion of failed rendezvous cases (17, 27). This open-ended approach to EUSBD (i.e., including both rendezvous and EUS-CDS), results in comparatively higher success rates than that of an EUSBD series, hence limiting their approach to just rendezvous (26). It is important to know that choledochoantrostomy, described by Artifon et al. (28), is a new technique that is useful for those patients with duodenal bulb infiltration, and may be a new and feasible tool as a variant of choledochoduodenostomy.