The first-line treatment for mild and moderate acute cholecystitis is early laparoscopic cholecystectomy . In patients at high surgical risk, gallbladder drainage is a frequently chosen alternative , and percutaneous transhepatic gallbladder drainage (PTGBD) is the most commonly employed method. However, it has some drawbacks: it is uncomfortable and painful, drainage dislodgment takes place in up to 20% of patients , it is unsuitable for patients with coagulopathy or with massive ascites, and the rate of recurrence in patients who do not undergo cholecystectomy ranges between 22% and 47% .
Endoscopic transpapillary gallbladder drainage (ETGBD) and endoscopic ultrasonography-guided transmural gallbladder drainage (EUS-GBD) are promising alternative treatments for patients at high surgical risk. ETGBD solves 90% of cases of acute cholecystitis  and provides definitive drainage in patients unfit for surgery. Its main drawbacks are the technical difficulty of advancing a guidewire through the cystic duct, which accounts for the <90% technical success described in some series , and the small caliber of the stents placed through the cystic duct, which are prone to obstruction. EUS-GBD is the most recent alternative to percutaneous cholecystostomy. A recent study showed that EUS-GBD and PTGBD have similar rates of technical (97% vs. 97%) and clinical (100% vs. 96%) success, without a significant difference in complication rates (7% vs. 3%) .
The first stents used for EUS-GBD were self-expandable metal stents (SEMS), initially designed for intraluminal deployment. Therefore, they might be more prone to bile leakage and migration, as they do not provide lumen anchorage. Lumen-apposing metal stents (LAMS) were specifically designed to anchor to hollow viscera, solving the limitations of conventional SEMS in this setting , although concerns about adverse events have been raised. A recent cautionary report of an ongoing trial (NCT02685865) evaluating endoscopic drainage of walled-off necrosis  described significant delayed adverse events in 6 patients after LAMS were deployed in 12. On the other hand, studies evaluating EUS-GBD have not shown such high rates of adverse events [10,11] and long-term stent indwelling might reduce the risk of recurrence , but data beyond the year of follow up are still scarce. Thus, the convenience of leaving the LAMS in situ in EUS-GBD performed in non-surgical candidates is still a matter of debate.
Our study aimed to analyze the long-term (>1 year) outcomes of subjects who underwent EUS-GBD employing a LAMS, including new biliary events, hospital admissions, and all-cause mortality.
Patients and methods
The present study was a single-center retrospective case series evaluating the long-term outcomes of patients who underwent EUS-GBD for acute cholecystitis. It was approved by the local institutional review board (IRB). All authors had access to the study data and have reviewed and approved the final manuscript.
Selection of patients
Patients were prospectively enrolled in an IRB-approved LAMS registry, including all LAMS deployed in a single academic tertiary care center. All patients or their legal representatives provided written informed consent. Those who underwent EUS-GBD for acute cholecystitis between May 2011 and November 2015 were eligible to participate in this study. Patients with an indwell time <12 months were excluded. Short-term outcomes of some of these patients have been reported previously [10-13].
All procedures were performed by 2 expert endoscopists (MPM, CdlSH) in the endoscopy suite. All patients were sedated by intravenous administration of midazolam and propofol. The gallbladder was imaged under EUS from the antrum or the duodenal bulb with a therapeutic echoendoscope  and then punctured with a 19-G needle (Expect; Boston Scientific), avoiding intervening vessels. After that, a 0.035-inch guidewire (Hydra Jagwire; Boston Scientific) was passed through the needle and coiled in the gallbladder. The LAMS (AXIOS, Boston Scientific) deployment technique depended on the type of stent chosen. If a conventional LAMS (AXIOS, Boston Scientific) was chosen, serial dilatation with 6-Fr cystotome (Cysto-Gastro-Set, Endo-flex) followed by a 4-mm biliary balloon (Hurricane, Boston Scientific) was performed prior to the insertion of the stent under EUS and fluoroscopic guidance. In case of an electrocautery-enhanced (hot) LAMS (AXIOS, Boston Scientific) the stent was deployed directly over the guidewire. Hot LAMS also permits a freehand technique, whereby direct access to the target is achieved without prior needle puncture. The LAMS used were 10 or 15 mm in diameter. Coaxial double-pigtail plastic stent (Boston Scientific) insertion and dilation within the stent lumen with a 10- or 15-mm balloon dilator (Controlled Radial Expansion balloon dilator; Boston Scientific) to achieve rapid deployment and prevent potential dislodgement secondary to therapeutic maneuvers through it were performed at the discretion of the endoscopist.
Acute cholecystitis was diagnosed according to the Tokyo guidelines criteria, based on a combination of clinical symptoms (fever, right upper quadrant pain, positive Murphy’s sign), laboratory data (high level of serum C-reactive protein, leukocytosis) and imaging findings (US, EUS or CT) . Cholecystectomy was dismissed in all cases because of the patients’ advanced age and poor physical status (class ≥III on the American Society of Anesthesiologists’ Physical Status classification). None of the patients improved after 24-72 h of conservative management with intravenous antibiotics, fluid replacement and bowel rest.
Complications were defined as any procedure-related event appearing during or after the procedure; they were described according to their nature and graded for severity according to the American Society for Gastrointestinal Endoscopy lexicon’s severity grading system .
Causes of admission during follow up were categorized according to the diagnoses stated in the discharge reports, codified according to the International Classification of Disease 9-Clinical Modification (ICD9-CM). End of follow up was defined as the occurrence of death, relapse, discharge from the outpatient clinic, or, in the absence of any of these, the last outpatient clinic medical visit or the last telephone follow up.
To assess the long-term clinical outcomes of EUS-GBD with LAMS in definitive non-operative candidates.
Data regarding baseline demographics and diagnosis, endoscopic procedure, adverse events, migration, stent retrieval, and mortality were retrieved from the prospective LAMS registry available in our center. Fifteen patients were followed in our institution. The remaining 7 patients belong to another 4 institutions. In addition, all-cause emergency room visits and hospital admissions were retrieved from medical electronic records, which included all notes made by physicians on outpatient visits, as well as reports of analysis, cultures, imaging studies, and all other procedures performed. Discharge reports after every hospital admission were also included. All patients were contacted by phone to ensure that no procedures or admissions at other centers were missed. Patients’ data were collected into a newly created database. In case of unclear or contradictory statements in the medical record, a consensus decision was made.
The analysis was performed with Stata (StataCorp. 2013. College Station, TX). Categorical variables were represented as percentages. Continuous variables with a normal distribution were presented as mean and standard deviation and those without a normal distribution were summarized as median and interquartile range (IQR).
A total of 47 patients with EUS-GBD employing a LAMS were identified. Twenty-five of them were excluded for having an indwell time under 12 months, as shown in Fig. 1. Stent dysfunction included 1 partial gastric outlet obstruction and an angulated stent resulting in relapse of cholecystitis. One of the 12 deaths presenting in the first year of follow up was related to the procedure (hemoperitoneum). Thus, 22 patients were included in the final analysis: 14 (63.6%) male and 8 (36.4%) female with a median age of 88.3 years (IQR: 82.6-92.7; range 71.3-97.5). Fourteen (63.6%) patients had undergone ERCP, 11 of them in the same procedure as when the LAMS was deployed.
Figure 1 Flowchart showing patient selection
EUS, endoscopic ultrasound; LAMS, lumen-apposing metal stent.
All procedures are summarized in Table 1. Transgastric access was chosen in 15/22 patients while coaxial stents were placed in only 3/22 subjects.
Table 1 Deployment intervention
Median follow up was 24.4 months (IQR: 18.2-42.4; range 12.3-62.4). All emergency room visits, hospital admissions and causes of death are summarized in Table 2. During long-term follow up, 12 patients (54.5%) visited the emergency room 34 times with a median of 1 visit/patient (IQR: 0-3; range 0-7). A total of 36 hospital admissions were required, with a median of 1 admission/patient (IQR: 0-3; range 0-9). Among the 22 patients included, 14 (63.6%) patients died during follow up: one of them of pancreatic cancer progression, while the rest died from non-biliary causes.
Table 2 Causes of death, hospital admissions and emergency room visits
In 15 patients (68%), imaging at least 3 months after deployment was available, confirming the presence of the LAMS in 100%. In the remaining 7 patients there was no image available from at least 3 months after deployment. Median time from LAMS deployment to the last available imaging procedure was 607 days (IQR: 463-938; range 186-1566). Fig. 2A and B present in situ LAMS in CT scans performed during hospital admissions.
Figure 2 (A) Computed tomography (CT) scan performed for obstructive colonic cancer showing a lumen-apposing metal stent (LAMS) with an indwell time of 4 years. (B) CT scan of a patient with a LAMS indwell time of 27 months
Long-term endoscopic follow up was available in 3 patients. One patient with a cholecystoduodenostomy was evaluated 31 months after deployment and found to have a patent stent (Fig. 3A). One patient with a cholecystogastrostomy, in an esophagogastroduodenoscopy performed 42 months after the deployment, exhibited a significant overgrowth in the LAMS causing obstruction of the stent (Fig. 3B). Another patient with a cholecystogastrostomy showed a patent fistula in the last endoscopic follow up, but not the stent (Fig. 4A). An abdominal plain film (Fig. 4B) confirmed the suspected buried-stent syndrome, confirming the presence of the LAMS in the right upper quadrant. Table 3 displays data regarding migration and patency evaluation.
Figure 3 (A) Patent cholecystoduodenostomy employing a 10×10 mm lumen-apposing metal stent (LAMS) 31 months after deployment. (B) Cholecystogastrostomy showing overgrowth 42 months after LAMS deployment
Figure 4 Cholecystogastrostomy with a 15×10 mm lumen-apposing metal stent 61.5 months after deployment, showing a patent fistula with a completely buried stent (A), confirmed by its presence on the abdominal X-ray film (B)
Table 3 Radiologic and endoscopic data assessing migration and patency
Biliary and gastrointestinal events during follow up
Two of the 22 patients (9%) required new hospital admissions for biliary disease, but just 1 for gallstone-related disease (4.5%): a patient presenting 2 episodes of moderately acute cholangitis with choledocholithiasis. The other presented obstructive jaundice and moderately acute cholangitis related to pancreatic malignancy. Neither recurrences of cholecystitis nor cases of gastric outlet obstruction or gallbladder sump syndrome were observed during follow up.
Six episodes of upper gastrointestinal bleeding were diagnosed in 2 patients, none of them related to the LAMS. One patient presented peptic erosive esophagitis and the other presented 5 episodes of acute variceal bleeding and portal hypertensive gastropathy.
Our study presents the first cohort published to date analyzing long-term outcomes of EUS-GBD with indwelling LAMS. Patients with an indwell time ≥12 months were identified from a prospective LAMS registry that included baseline demographics and endoscopic procedure data. Follow up was documented with medical records, endoscopy, imaging and phone contact.
It is important to highlight the complexity of the management of this specific population. This is shown by the fact that 25.5% of the initial cohort died before completing the first year of follow up. The difficulty of the procedure, even in expert hands also poses a risk of dysfunction, migration or even surgery. Nevertheless, acute cholecystitis presents a high mortality in this group of patients  and EUS-GBD aims to both control the acute infection and reduce the risk of relapse in a single procedure.
Focusing on our cohort, we observed a relevant effect on the number of gallstone-related admissions, with only one patient (4.5%) requiring 2 admissions for acute cholangitis during a median follow up of 24.4 months. Although our cohort was small, this proportion is similar to the 4% readmission rates identified in patients undergoing cholecystectomy in large retrospective studies with a follow up of 2 years . Our findings also resemble those shown in the largest study published to date, by Choi et al, which included 56 patients with a shorter follow up (median 275 days, range 40-1185), presenting recurrent acute cholecystitis in 2 patients (3.6%) . It should be taken into account that patients with cholecystitis who do not undergo cholecystectomy present rates of readmission for gallstone-related disease of 38% in the subsequent 2 years .
Adverse events are a significant concern in the long-term deployment of LAMS. Currently available evidence is drawn mainly from studies evaluating the management of pancreatic fluid collections. A recently published report of an ongoing trial (NCT02685865) that evaluated LAMS for walled-off necrosis drainage reported severe delayed gastrointestinal bleeding in 3 patients, buried LAMS syndrome in 2 patients and biliary stricture in 1 of the 12 patients included , but previous published series and ongoing registries have shown significantly lower rates of adverse events, ranging from 5-15% [18-21]. The possibility of extrapolating these results to EUS-GBD is yet to be determined, as pancreatic fluid collections are pathological cavities with an inflammatory cause and their size collapses after treatment. No LAMS-related adverse events were observed in our cohort of patients after the first year of follow up, although there were early adverse events, including 2 cases of stent migration, 2 of stent dysfunction and 1 fatal bleeding. The study by Choi et al identified no adverse events beyond the first year . A multicenter retrospective study by Walter et al, which included 15 patients whose EUS-GBD was left in situ with a mean follow up of 364 days (standard deviation: 67 days) , reported 4 procedure-related adverse events (13%), although the time to adverse events was not reported. We hypothesize that late migration and gastrointestinal bleeding are infrequent because of tissue overgrowth.
This study has significant limitations which should not be overlooked. Firstly, its retrospective design might have underestimated the number of adverse events. We tried to correct for this with a thorough search of medical records and phone contact with patients or their relatives. Secondly, long-term endoscopic or imaging studies were not available in all patients and the imaging follow up was shorter than the clinical one. Thus, the number of migrations might be underestimated. Nonetheless, migration of LAMS after EUS-GBD seems to be frequently symptomatic (all identified migrations in our cohort were symptomatic), although in a study by Choi et al there were 2 asymptomatic migrations  in the first year after deployment. Thirdly, the single-center design of the study, with all procedures performed by 2 expert endoscopists, might not allow the extrapolation of results to other centers where endoscopists may have varying levels of expertise in EUS-GBD or limited experience in the use of LAMS.
In summary, our study shows that long term indwell of EUS-GBD with LAMS provides a definitive drainage of the gallbladder, reducing the risk of further biliary events. Moreover, leaving the LAMS in place avoided a further invasive procedure in elderly patients with high surgical risk. We observed no delayed LAMS-related adverse events after a thorough assessment of all-cause emergency room visits, admissions and mortality. Thus, we suggest permanent EUS-GBD should be considered in patients with high surgical risk who present with moderate/severe acute cholecystitis, although large comparative studies are needed to confirm these promising results.
What is already known:
The first-line treatment for acute cholecystitis is laparoscopic cholecystectomy
Endoscopic ultrasound-guided gallbladder drainage, when performed by skilled endoscopists, is an effective alternative treatment for acute cholecystitis in patients unfit for surgery
Long-term results beyond 1 year of follow up are still scarce
What the new findings are:
Long-term indwell of endoscopic ultrasound-guided gallbladder drainage using lumen-apposing metal stents (LAMS) reduces the risk of further biliary events in patients who do not undergo cholecystectomy
No delayed LAMS-related adverse events were identified after a median follow up of over 2 years