Cross-sectional imaging should be performed prior to initiation of endoscopic debridement to evaluate the characteristics of the collection (i.e., size, shape, wall thickness), discern any intervening vasculature, and determine the relationship to the lumen wall. The computed tomography (CT) appearance of WON can vary. Most commonly pancreatic necrosis is detected radiographically on contrast-enhanced CT by the presence of non-enhancement of the pancreatic parenchyma and surrounding structures. However, it may appear homogenous, similar to that of an acute pseudocyst, as underlying solid debris is frequently indistinct on CT. This may lead one to embark on standard pseudocyst drainage methods that inadequately remove the underlying solid material potentially resulting in serious infection.

In patients with evidence of classic WON on cross-sectional imaging, it is necessary to discern the infection status of the necrotic cavity, as clinical signs of infection (i.e., leukocytosis, fever) are frequently insufficient. Percutaneous fine needle aspiration (FNA) may be used to determine the infection status of the necrosis, though this is not often necessary or performed. Urgent endoscopic drainage and debridement mandated in patients with microbiologic evidence of infected necrosis.

The indications and timing of drainage of sterile pancreatic necrosis are more controversial. Generally, pancreatic necrosis is not amenable to endoscopic drainage until 4–6 weeks after onset of pancreatitis, allowing the process time to organize and encapsulate. As a general rule, endoscopic drainage and debridement should be delayed as long as possible in patients demonstrating clinical stability. The most common indications for drainage of sterile necrosis include the following: (1) persistent abdominal pain; (2) evidence of gastric outlet obstruction (clinical or radiologic) (Fig. 12.2); (3) biliary obstruction; or (4) failure to thrive (ongoing systemic illness, anorexia, and weight loss). The appearance (i.e., size, location) of the necrotic collection on cross-sectional imaging may not be indicative of the patient’s clinical status and that finding alone is not an indication for intervention.

Basic pre-procedural tasks are also vitally important. The INR and platelet count should be obtained and corrected, as necessary. Pre-procedural, broad-spectrum antibiotics should be administered in patients not already receiving them. Intravenous penicillins (i.e., piperacillin/tazobactam), quinolones (i.e., levofloxacin), or carbapenems (i.e., meropenem) are considered recommended agents. Antibiotic therapy should be tailored based on microbiologic cultures obtained during the procedure. Of note, we perform all direct endoscopic necrosectomy (DEN) procedures with general anesthesia given the patient acuity, length of the procedure, and higher risk for intraprocedural adverse events. Given the risk of air embolism with conventional endoscopic insufflation, CO₂ insufflation is routinely used during endoscopic necrosectomy.

Careful review of cross-sectional imaging is vital. Coronal CT images can be very useful and often provide complementary information to standard axial images. Understanding the degree of necrosis, including extension into the paracolic gutters and communications between cavities, will direct the therapeutic plan and promote efficient therapy. Patients may appear to have multiple separate cavities, though these are generally extensions of the same area of necrosis. Non-dependent locules of air are also frequently seen within collections, though this finding in no way implies infection with a gas-forming organism but instead commonly represents a fistulous connection to the gastrointestinal lumen. In certain clinical scenarios, this fistulous tract can be used for transmural entry into the cavity to provide egress or to facilitate debridement as described below.

The endoscopic approach to the management of WON is predicated on evacuation of solid debris from the necrotic cavity. An initial transmural puncture through the gastric or duodenal wall is necessary to facilitate access to the collection, and to drain liquefied material. For WON located within or adjacent to the mid-body or tail of the pancreas, a transgastric route is preferable, while a transduodenal puncture may be necessary for collections confined to the pancreatic head. Non-endoscopic ultrasound (EUS)-guided punctures using fluoroscopic guidance can be successfully performed (>95%) with low adverse events (<5%) when an obvious luminal protrusion is observed endoscopically [4], though most experts agree that EUS guidance is preferred if available. EUS provides precise targeting of the lesion, potentially mitigates inadvertent damage to adjacent vasculature, and allows real-time assessment of the extent and volume of the necrotic cavity [5].

A variety of endoscopic accessories can be used to perform the transmural puncture, including electrocautery-based instruments such as needle knives and specialized fistulotomy devices (Cystotome, Cook Endoscopy, Winston-Salem, NC), and non-cautery accessories such as EUS-FNA needles. A newer stent with electrocautery-enhanced delivery system (described below) has recently become available. Entry into the cavity is confirmed by extravasation of cyst contents within the lumen during puncture, aspiration of cyst fluid through a needle, and/or the injection of radiopaque contrast in the cavity under fluoroscopy. One advantage of FNA or other aspiration needles is the ability for guidewire passage through the needle into the cavity using the Seldinger technique, following drainage of liquefied contents. We typically utilize a specialized 19-G FNA needle (EchoTip Ultra HD Ultrasound Access Needle, Cook Endoscopy) designed for such procedures.

After puncture and drainage, a guidewire is passed into the cavity and the transmural tract is dilated with a standard dilating balloon to a diameter of ≥15 mm. It is important to pass an ample length of guidewire into the collection in preparation for subsequent stent placement. If guidewire loss occurs inadvertently, it may be challenging to re-access the cavity and may increase the risk for adverse events while attempting to do so, even despite prior tract dilation.

Once a stable guidewire is placed, some endoscopists elect to place one (or more) double-pigtail plastic stents prior to performing necrosectomy. This technique is especially important with transgastric access, as it may be challenging to identify the puncture site among gastric folds. When using plastic stents, we recommend using two 10-Fr double-pigtail stents with a length of 3–5 cm to minimize the risk of stent migration into or out of the cavity, and stent impaction causing trauma to the lumen or cavity wall. Care must be taken not to deploy the entire plastic stent within the collection. We routinely place an endoscopically visible indelible mark at the midpoint of the stent prior to placement to guard against this situation. An alternative option is to place a large-bore (16–23 mm mid-body diameter) self-expandable metal stent (SEMS) across the dilated gastric or duodenal wall [6–10]. Transmural placement of large-diameter, covered esophageal SEMS facilitates subsequent DEN procedures and avoids the need for repeated balloon dilation prior to each debridement [7]. The use of esophageal SEMS is somewhat limited as the shortest SEMS lengths are 6–7 cm, resulting in excessive stent length within the gastrointestinal lumen or the necrotic cavity. The excess length can be trimmed using argon plasma coagulation to prevent impaction against the lumen or cavity wall, though care should be taken to minimize disruption of the SEMS interstices. An alternative option is to place a double-pigtail plastic stent within the deployed SEMS. The double-pigtail stent serves two purposes: (1) It acts as a bumper between the stent flange and the lumen/cavity wall, and (2) it prevents loose necrotic debris from obstructing the SEMS.

The recent development and FDA-approval of a lumen-apposing metal stent (LAMS) (AXIOS; Boston Scientific, Marlboro, MA) serves to overcome many of the limitations of our current drainage options (Fig. 12.3). These stents are available with mid-body luminal diameters of 10 and 15 mm and external flange diameters of 21 and 24 mm, with an overall stent length of ~1 cm. Two AXIOS stent delivery systems are available, one which requires placement via the standard technique (i.e., 1-puncture, 2-guidewire placement, 3-tract dilation, 4-stent deployment) and the other which is fitted with an electrocautery tip designed for simultaneous puncture and tract dilation followed by stent deployment without the need of guidewire placement. This short-length LAMS is ideal for DEN, as the 15 mm stent diameter accommodates repeated endoscope entry and exit necessary for adequate debridement, while providing apposition between the gastric wall and cavity wall [11, 12].

Spontaneous fistulous tracts can also be utilized to access, drain, and debride a necrotic collection [13]. Puncture and subsequent drainage is followed by balloon dilation of the tract to a diameter ≥15 mm. Dilation allows egress of the remaining liquid and solid material and facilitates DEN. Adjunctive transpapillary stenting can be considered; however, this technique only allows drainage of additional simple peripancreatic fluid and is inadequate as a conduit for removal of pancreatic debris.

A variety of techniques have been described for removal of solid debris from necrotic collections. Some endoscopists advocate for irrigation of the cavity through placement of a 7-Fr nasocystic irrigation tube within the collection, adjacent to the transmural stents, using one or more transmural exit sites [14, 15]. Up to 200 cc of normal saline (±3% hydrogen peroxide) is forcefully and rapidly infused via the tube every 2–4 h initially to lavage debris from the cavity. Many providers have abandoned the use of nasocystic irrigation tubes for current clinical practice, mainly due to patient intolerance [16]. A variation of this technique includes the combination of endoscopic and percutaneous therapy, termed “dual-modality therapy” [17]. This technique involves percutaneous drain placement, in lieu of a nasocystic tube, and endoscopic transmural drainage with irrigation via the percutaneous catheter and egress through the transmural tracts to promote ongoing debridement of the necrotic tissue [17].

In the majority of cases we advocate for DEN if clinically appropriate. DEN is a time-intensive procedure, which requires passage of an endoscope transmurally into the collection. Both diagnostic and therapeutic endoscopes can be used, each with inherent advantages and disadvantages. Diagnostic endoscopes can be advantageous when maximal flexibility is desired. Therapeutic endoscopes offer a high-flow water jet which is particularly helpful in fragmentation and irrigation of necrotic tissue. The use of hydrogen peroxide lavage may be effective in liquefying necrotic tissue during DEN, though evidence from comparative trials is lacking [18]. Endoscopic debridement can be performed utilizing a variety of standard endoscopic accessories (i.e., stone retrieval baskets, polypectomy snares, polyp-retrieval nets, grasping forceps, etc.). Once the necrotic tissue is grasped, it is withdrawn from the cavity and deposited in the lumen. Alternatively, large pieces may be cut into smaller pieces using snare electrocautery and flushed from the cavity; this may avoid repeated entry into and withdrawal from the cavity. The consistency of necrotic debris varies from patient to patient, ranging from smooth, solid debris that is densely adherent, to tissue that is loose and easily removed. Some endoscopists routinely perform DEN following the initial transmural puncture, while others advocate for initial drainage with debridement commencing during a second procedure. No data exist to suggest a benefit of one strategy over the other. The goal of each DEN procedure should be to remove as much necrotic tissue as possible.

Following DEN, patients can resume (or initiate) oral intake the day of the procedure, assuming no intraprocedural adverse events have occurred. Per-oral antibiotics are generally continued for several weeks and in most cases until complete resolution of the cavity. Repeat procedures are almost always necessary and frequently include stent exchange and additional debridement. Subsequent procedures can be scheduled if the initial necrosectomy is known to be incomplete [19], or performed as necessary based on clinical status and/or cross-sectional imaging findings. Debilitated patients requiring hospitalization may need more frequent procedures (every 1–2 days), while outpatients who continually improve may tolerate 1–2 weeks between interventions. The interval between debridements is often predicated on the clinical scenario in conjunction with logistical issues (i.e., inpatient/outpatient status, distance from treatment center, and availability of an advanced therapeutic endoscopic team). In patients with continued collections despite multiple interventions, endoscopic retrograde cholangiopancreatography (ERCP) should be considered for evaluation of an ongoing pancreatic duct disruption. As resolution ensues, external drains should be removed before internal drains to prevent formation of gastro-/enterocutaneous fistulae. Internal drains are then endoscopically removed after complete resolution of the collection.

Many case series have demonstrated the efficacy of DEN [16, 20–23]. Patients with WON are a heterogeneous group with notable variation in the following: (1) size of the necrotic collection, (2) total burden of necrosis, (3) the presence of paracolic gutter extension(s), (4) comorbid medical illnesses, and (5) time from onset of necrosis to intervention. Thus, comparison of outcomes between reported series remains challenging as definitions for outcomes vary substantially. Successful resolution can be defined as complete non-surgical resolution (including percutaneous drainage) or resolution due to flexible endoscopy alone [16].

Two recent systematic reviews, including 233 and 455 patients, respectively, have demonstrated complete resolution of pancreatic necrosis in 81% of patients using endoscopy alone [24, 25]. The mean number of procedures necessary for resolution was 4 in both studies, while the adverse event rates were 21 and 36%, respectively. Two large retrospective studies of DEN showed successful resolution in approximately 90% of patients with an adverse event rate of approximately 14% [16, 21].