According to the revised Atlanta criteria, walled-off pancreatic necrosis (WOPN) consists of collections derived from a necrotizing pancreatitis as the liquefied parenchyma matures into a contained collection [1]. Traditionally, interventions for WOPN have been surgical or percutaneous, and endoscopic management of these necrotic collections is a relative newcomer.

The first endoscopic transmural drainage of a WOPN was reported in 1996 by Baron et al. [2]. More recently, a limited number of large multicenter studies have demonstrated the efficacy and safety of endoscopic transmural necrosectomy for WOPN which now is accepted as first-line treatment for WOPN when indicated [3–5]. However, the indications and methodology described vary between studies, and the endoscopic techniques are complex and continuously evolving [6].

For these reasons, management of WOPN is an area of great interest and current research with as yet many unanswered questions. For example, it is unclear how aggressive endoscopists should be in the debridement of the necrosis and when and how frequently the debridement should be performed. Other concerns include the roles of mechanical necrosectomy versus irrigation, and which type of stents should be used [7, 8].

Recently, some case series have reported the initial experience using a new type of self-expanding metal stent (SEMS), specifically designed for endoscopic ultrasound (EUS)-guided transmural drainage of collections, defined as lumen-apposing metal stents (LAMSs). Stents of this kind could be used as ports for repeated sessions of necrosectomy [9].

To our knowledge, no study has evaluated endoscopic drainage using LAMS combined with sessions of energic flushing of the cavity for debridement of the debris, as in irrigation necrosectomy, avoiding the direct manipulation of adherent necrotic material as a direct endoscopic transmural necrosectomy.

The aim of this study was to evaluate, in terms of feasibility and safety, the outcome of a new variant of the endoscopic treatment for WOPN, using a LAMS and irrigation sessions through this stent.

Materials and methods

The study took place at Hospital Universitari de Bellvitge, a tertiary-care public institution in the Barcelona area (Catalonia, Spain). Patients and procedural data from September 2011 to August 2014 were prospectively recruited and analyzed retrospectively from a specific database. The following information was collected: demographic data, procedure details, clinical outcomes, adverse events, and follow-up after endoscopic treatment. The study concept was investigator initiated, and no financial support or free devices were received. Written informed consent was obtained from all patients. Our institutional review board approved the technique, deeming it to be an extension and variant of existing procedures. All procedures were performed by one experienced endoscopist.

Patient data

All patients with symptomatic WOPN were considered eligible for this study. Criteria for drainage were as specified by the Working Group of IAP [10]. The exclusion criteria were: imaging findings suggesting a pseudocyst, collections extended into the lower abdomen/pelvis or endoscopically inaccessible from the gastric cavity or duodenum, severe coagulopathy (prothrombin time >1.5 or platelet count <50 × 109/L), age younger than 18 years, and inability to give a written informed consent.

Material

All access sessions were performed under orotracheal intubation. For irrigation or removal stent sessions, only general anesthesia was used. Patients were placed on the left side or in a prone position. For access sessions, only seven procedures were carried out in a fluoroscopic room. The rest of the procedures were performed in the EUS room. In all access sessions of non-infection collections, prophylactic doses of intravenous antibiotics were administered before the procedure and 7 days following.

EUS-guided transmural drainage was performed with a therapeutic linear echoendoscope (GF-UCT140, Olympus Europa). Other sessions (irrigation, stent removal) were performed using a standard upper endoscope (GIF-Q145, Olympus). A fully covered LAMS was used in all cases (AXIOS or Hot AXIOS, 10 or 15 mm × 10 mm, Xlumena Inc., Mountain View, CA, USA). In six cases, double-pigtail plastic stent were placed through the LAMS (10F × 5 cm, Cook Endoscopy, NC, USA). In one case of massive bleeding inside the drained cavity, an 8.5F nasocystic catheter (Boston Sc., Natick, MA, USA) was placed within an LAMS and coaxial to a pigtail stent.

In four procedures, an LAMS was placed using a novel cautery-tipped stent delivery system called Hot AXIOS (Xlumena Inc.).

The irrigation endoscopic necrosectomy sessions were performed using an endoscopic flushing powerful pump (OFP, Olympus) that can irrigate saline via the instrumental channel of the endoscope.

Technique

Access session

Initial endoscopic access was obtained using EUS-guided transmural drainage. An LAMS was placed to create a wide stable access to the cavity.

  • First, a linear echoendoscope was used to visualize the extent of the collection and necrosis.

  • In nine cases, the cavity was punctured with a 19-gauge needle (Expect Flex, Boston Sc.), and an aspirate was obtained for analysis. A guidewire was inserted through the needle into the necrotic cavity under fluoroscopic or EUS guidance. The puncture site was dilated using a 6F cystotome (EndoFlex, Voerde, Germany) and a biliary (6 or 8 mm, Boston Sc.) or CRE (10–12 mm, Boston Sc.) dilating balloon.

  • In four cases, the puncture and ostomy were performed directly and at the same time using the Hot AXIOS cautery-tipped (as a cystotome) stent delivery system.

  • In all procedures, an AXIOS stent was placed over the guidewire and deployed under EUS guidance.

  • Then, one or more double pigtail stent(s) and/or a nasocystic drainage catheter were inserted. In a few cases, endoscopic necrosectomy was performed on the same day.

Irrigation sessions

  • After the access session, irrigation necrosectomy sessions were planned at intervals of 2–5 days until most of the non-adherent necrotic material was pushed out, and there was clinical improvement of the patient.

  • If necessary to gain a better view, a gastroscope was entered into the cavity, but in the majority of cases this was not necessary.

  • These lavage sessions were performed by flushing SSF (500–1500 mL) through the LAMS using a water-jet system. All of the SSF flushed was totally aspirated by suction until the cavity was empty, removing all the non-adherent necrotic debris from the gastric cavity.

  • As a final step, LAMS was removed after WOPN resolution and confirmed by imaging.

  • Initially, the first session was restricted to insertion of the LAMS (±double plastic pigtail) aspirating all the fluid content of the cavity, but with increased experience, irrigation sessions were carried out during the first session.

Examples of the techniques applied in the access and irrigation sessions are shown in Figs. 1 and 2, respectively.

Fig. 1
figure 1

Different images showing transluminal endoscopic ultrasound (EUS)-guided drainage of walled-off pancreatic necrosis (WOPN) with a lumen-apposing metal stent (LAMS). A Endosonography view showing a WOPN with abundant solid parts. A guidewire is seen as a hyperechogenic line through the gastric wall into the cavity. B Puncture of the collection using a 19G needle and a guidewire inserted into the cavity. C Access into the lesion with the insertion of cautery-tipped stent delivery system under EUS guidance. D Endoscopic view of the Hot AXIOS cautery tip. E Images of AXIOS stent deployment under EUS guidance. F EUS image of an AXIOS stent delivered. G Endoscopic view of the AXIOS stent deployed and well positioned, with abundant thick fluid drained into the stomach. H A pigtail inserted through the LAMS

Full size image
Fig. 2
figure 2

Sequence of endoscopic images showing an endoscopic necrosectomy session with irrigation technique using a high-flow water-jet system. A Endoscopic view across an occluded AXIOS stent with necrotic debris. B AXIOS lumen submerged underwater. C Necrotic fragments lie in the gastric cavity after being dragged in the necrosectomy session by irrigation. D A guidewire is inserted through the AXIOS stent. E A double-pigtail plastic stent within the AXIOS stent creating space between necrotic fragments and lumen stent, to avoid self-occlusion. F Endoscopic view of the 10F double pigtail placed within the AXIOS stent

Full size image

Post-procedure care and follow-up

  • All patients were monitored and admitted to our center for observation after the procedure.

  • WOPN resolution was assessed by MDCT (multidetector computed tomography) or EUS generally 4–8 weeks after initial transmural drainage. If complete resolution was achieved, all stents were removed.

  • However, if the collection was not resolved, stent removal was deferred and new imaging procedure was scheduled for 2–4 weeks later.

  • In patients with suspicion of possible pancreatic duct disruption, stent removal was deferred and a cholangio-magnetic resonance imaging (MRI) was proposed to evaluate the need for a therapeutic ERCP.

  • After the removal of the stents, follow-up visits with an MDCT scan were scheduled for each case at between 2 and 6 months.

Outcome measures: definitions of events

Procedure time was calculated as the time from the insertion of the first scope to the withdrawal of the last.

The term WOPN is defined, according to the revision of the Atlanta classification, by the presence of necrotic tissue in the collections, based on a history of necrotizing pancreatitis [1].

Technical success was defined as the correct placement of the stent. Clinical success was defined as a decrease in the size of the collection by ≥50 % at 4- to 6-week follow-up imaging, together with resolution of symptoms.

Adverse events were defined and graded according to the standard lexicon for endoscopic adverse events.

Irrigation necrosectomy was defined as vigorous irrigation of the WOPN cavity with a large amount of SSF followed by aspiration of the saline, without using mechanical tools for the debridement.

Recurrence was defined as symptomatic collection diagnosed on imaging procedure following initial treatment success.

Results

Patient characteristics

Twelve patients (52.6 ± 14.4 years) with 13 WOPN collections (median size 12.4 ± 2.94 cm; range 8–18 cm) underwent endoscopic treatment with the reported technique.

Of the 12 patients, one was in very poor health and five had severe systemic disease, according to the American Society of Anesthesiologists (ASA) physical status classification.

All patients had a previous acute necrotizing pancreatitis which, after 4 weeks, developed partially liquefied collections with necrotic debris. These collections were well identified by the imaging procedures (MDCT or MRI). Indications for endoscopic intervention (transmural drainage ad necrosectomy) in these necrotic collections were abdominal pain, infection, food or biliary obstruction, disconnected pancreatic duct syndrome, and severe stricture of splenic vessels with high risk of bleeding.

The baseline characteristics and demographics of each case with procedural outcomes are summarized in Table 1.

Table 1 Patient demographics and procedural outcomes
Full size table

Outcomes

All stents were successfully positioned in all cases, which means a technical success rate of 100 % was achieved. Resolution of the lesions was achieved in 100 % of cases after a median of three endoscopy sessions per patient (range 2–8).

The median length of hospital stay was 15.9 days. All patients experienced immediate symptom relief after endoscopic treatment.

General outcomes are given in Table 2.

Table 2 General outcomes
Full size table

Endoscopic procedures: observations

Median procedure time of the access session was 31 ± 10.16 min. The duration of irrigation sessions was usually limited to 30 min. A minimum of 48 endoscopic procedures was performed. In the case of Hot AXIOS, the median procedure time was 21.7 min ± 9.3 (9–28) versus 35.4 min ± 8.2 (20–47) in cases with device changes (p < 0.05).

All stents removed were easily retrievable after the resolution of the lesions. Mean time to stent retrieval was 9 ± 3.4 weeks.

A summary of all sessions done is presented in Table 3.

Table 3 Endoscopy sessions per patient
Full size table

Follow-up: recurrence

Patients were followed up prospectively after stent retrieval for a mean time of 13 ± 11.4 months. Only one patient presented a recurrence 12 months after stent removal, with a suspected disrupted pancreatic duct.

Adverse events

No adverse events (AE) were described during the procedures or 24 h after. But four patients experienced adverse events between the procedures. Two patients had a severe AE; in both cases, there were bleeding requiring transfusion, therapeutic arteriography, and endoscopic treatment, thereby increasing the number of irrigation sessions. Two cases had infection successfully treated with antibiotics and with endoscopy, and both were deemed to be moderate AE. No late AE occurred when observing all patients a minimum of 40 days after the initial procedure.

There was no procedure-related mortality.

Discussion

The present study offers the results of a new variant of endoscopic transmural necrosectomy in 12 consecutive patients with carefully selected WOPN lesions with a technical and clinical success rate of 100 % and faster sessions of transmural drainage. Furthermore, this methodology was found to be safe without need for surgery and with no mortalities. The technique is a combination of irrigation technique (without mechanical devices) and the use of specifically designed stents for transmural drainage guided by endosonography.

First of all, it is important to clarify that correct categorization of pancreatic fluid collection is the first step in its management. In some cases, MDCT cannot reliably detect necrotic debris within a pancreatic collection and many WOPN can be referred as pseudocysts. This is important, because endoscopic treatment outcomes are directly related to the type of pancreatic fluid collection being treated (90 % for pseudocyst, 65 % WOPN). For these reasons, it has become increasingly apparent that WOPN has been erroneously misclassified as a pseudocyst and inappropriately treated with transmural plastic stenting alone [11]. These necrotic collections contain highly dense solid debris and require a larger fistulous opening, or more than one tract, for effective drainage. In addition to an endoscopic transmural drainage, these collections often require an endoscopic transmural necrosectomy for debridement of solid debris, and they represent a true natural orifice transluminal endoscopic procedure.

Baron et al. in their initial experience with endoscopic treatment of WOPN used a mixed technique of transmural drainage with irrigation through a nasocystic drain [2], but later made recourse to more aggressive treatment, mechanical endoscopic necrosectomy. Today this is accepted as the most frequently indicated endoscopic option for WOPN.

The available case series and recent multicenter studies show that direct mechanical endoscopic necrosectomy is associated with high morbidity (15–30 %) and mortality (2–11 %), is labor intensive, is resource- and time-consuming, and is lacking in technique-specific devices [3–6]. In a recent systematic review that included 283 cases of pancreatic necrosis treated with EUS-guided necrosectomy, a median of five (1–35) sessions was required to achieve total resolution. A mean clinical success rate of 88 % and mean complication rate of 28 % were reported [12]. These aspects limit the practicability of direct endoscopic necrosectomy and point up the need for simplification.

Most studies of peroral endoscopic necrosectomy use plastic stents. It is well known that these stents have several limitations in the drainage of WOPN: a limited diameter, a high migration rate, and high occlusion rate. Placement of multiple stents can be tedious, and it is labor intensive and time-consuming. And there is no anchorage force in cases of non-adherent collections, with a high risk of leakage.

In the last 5 years, fully covered SEMS have been adopted for the drainage of pancreatic collections and, more recently, for the endoscopic treatment of WOPN [13, 14]. At the same time, new stents such as LAMS designed specifically for EUS-guided transluminal drainage have appeared, and they can be used as an access port for necrosectomy [15, 16]. As in our previous published report, in the present study we used the same type of stent with an anchoring design that offered a low stent migration rate, with only one migration of 36 of access and irrigation sessions [16].

Our cohort consisted of patients with WOPN (58 % infected), candidates for endoscopic treatment, in whom we planned to perform an initial transmural drainage plus vigorous irrigation sessions using specific transmural stents. This initial experience suggests that mechanical endoscopic necrosectomy can be avoided and be replaced to irrigation endoscopic necrosectomy, which is faster (30 min per session, 20 min using special devices such as Hot AXIOS), with fewer sessions (median of three instead of 5–7), and less prone to severe AEs (16 %) such as air embolism, severe bleeding, and perforation during the sessions, yet with comparable outcomes [6]. With increasing experience, we noted that it was better to directly insert a double pigtail within the LAMS during the first access session, in order to avoid external migration, occlusion of LAMS lumen with necrotic fragments (after an initial endoscopic necrosectomy, all fluid is evacuated and the cavity lumen is collapsed), or bleeding between the sessions due to trauma caused by the internal stent end.

Our group is a multidisciplinary team. In recent years, we have been working toward improvement of the management of local complications secondary to severe acute pancreatitis. After a critical review of 143 patients, we realized that aggressive surgical management was associated with a high mortality rate (postoperative mortality of 25 %), and minimally invasive options such as endoscopic treatments began to seem more attractive [17]. In seeking to improve our results, we compared data from our last consecutive patients, treated with direct mechanical endoscopic necrosectomy and conventional lumen stents (plastic pigtail alone or biliary SEMS), with the results of the present study. In doing so, we detected better clinical success, less time-consuming procedure, less number of sessions, and greater safety with less severe complications and no mortalities (Table 4).

Table 4 Comparison data from our unit with recent previous cases using mechanical devices and non-specifically designed stents
Full size table

Furthermore, comparing our results with the major multicenter studies (n > 55) reporting endoscopic transmural necrosectomy which only included necrotic collections, we found a higher rate of clinical success, a lower median number of sessions, lower rates of severe AE, and no mortalities (Table 5) [4, 18, 19].

Table 5 Major studies reporting endoscopic transmural necrosectomy of WOPN
Full size table

As noted in the literature, the most common complications associated with endoscopic necrosectomy are bleeding, perforation, infection, stent migration, and air embolisms. Bleeding usually occurs during balloon dilation of the transluminal tract or during the removal of necrotic material [6]. In our study, two serious AEs were described (16 %). Both were bleeding occurring between endoscopic necrosectomy sessions and not during stent placement. Curiously, these AEs occurred in two cases of LAMS stent alone (without plastic pigtail within). This suggests that the cause of the bleeding might be a vessel injury from the inner stent end. For this reason, all subsequent cases received an LAMS plus a pigtail stent, and no more bleeding events occurred. Nor were air embolisms or perforation described, perhaps as a result of having avoided exploration or manipulation of the cavity with the scope.

Limitations

Our study has some limitations that require further comment. The sample size was relatively small, although the number of total procedures was high (n = 47). As the study was carried out in a tertiary center with experienced interventional endoscopists, this might limit its generalizability to some extent. More importantly, this study was not randomized. A study comparing a mechanical endoscopic necrosectomy with this irrigation necrosectomy through an LAMS would require more patients and ideally would be a multicenter study. Additionally, we did not have long-term patient follow-up of all the cases, and a cost analysis was not carried out.

However, the good results from this technique have encouraged us to do a longer follow-up including more patients. We believe that the validity of our results is supported by a rigorous methodology; the data were prospectively collected, and the material used was standardized.

Conclusions

In conclusion, as shown by this pilot study, irrigation endoscopic necrosectomy by only flushing large amounts of saline through the LAMS, without any mechanical debridement, helps to simplify the technique, is faster and more effective, and yields excellent outcomes, without the need for surgery. Nevertheless, even though these clinical results are promising, larger prospective and randomized studies to validate and confirm these findings are in order.