A 72-year-old physically active retired family physician presented to our institution for evaluation and management of newly diagnosed pancreatic cancer. The patient was seen by his gastroenterologist 2 weeks before presentation with symptoms of jaundice, weight loss, and steatorrhea, and he underwent common bile duct (CBD) stent placement for obstructive jaundice at that time. Past medical history is notable for esophageal cancer treated with three-hole minimally invasive esophagectomy in 2008. Pathology at that time revealed moderately differentiated adenocarcinoma involving both esophageal and gastric mucosa, with 12 out of 21 examined lymph nodes positive for tumor. All surgical margins were negative for residual tumor. The patient subsequently underwent chemoradiation due to lymph node involvement. The surgical history also included a post-esophagectomy laparoscopic diaphragmatic hernia repair in 2011. Notably, the patient had smoked cigars for the past 15 years.

On presentation to our surgical clinic, scleral icterus and mild jaundice were noted. Laboratory studies revealed hyperbilirubinemia and elevated liver function tests. Both CA 19-9 and carcinoembryonic antigen levels were within normal range. Computed tomography (CT) scan revealed a pancreatic neck mass measuring 2.2 × 1.3 cm with obstruction of the CBD and main pancreatic duct (MPD) as well as possible superior mesenteric vein (SMV) involvement. Significant gallbladder distention secondary to biliary obstruction was also noted (Fig. 1). Endoscopic ultrasound confirmed the presence of a pancreatic head and neck mass obstructing and causing dilation of the MPD and CBD. Fine-needle biopsy of the mass revealed a poorly differentiated adenocarcinoma. Given the patient's prior history of esophagectomy, a CT angiogram was performed and confirmed blood supply to the gastric conduit via the right gastric artery, right gastroepiploic artery, and gastroduodenal artery (GDA; Fig. 2). Knowing that the right gastric artery was patent, a PPPD was planned to resect the pancreatic neck mass with the intent to sacrifice the GDA. The patient refused preoperative chemotherapy.

At the time of surgery, no dissemination was noted. A majority of the stomach was in the chest and not seen in the operative field. The pylorus was identified just below the diaphragm, as well as the first and second parts of the duodenum. Dissection of the duodenum off of the anterior aspect of the pancreatic head and neck was difficult because of the altered anatomy and the gastric pullup. The duodenum was carefully divided at least 3 cm beyond the pylorus, ensuring that the right gastric and right gastroepiploic arteries were left intact. The GDA was identified, suture ligated, and divided at its takeoff from the common hepatic artery. The proximal jejunum was divided about 20 cm below the ligament of Treitz. The duodenojejunal junction was mobilized, and the specimen was carefully separated from the adjacent mesenteric vessels. The pancreatic neck was divided to the left of SMV-portal vein confluence. The pancreatic head and neck were meticulously dissected from the right lateral aspect of the SMV and portal vein, and the specimen was separated from the visceral vessels. Typical end-to-side pancreatojejunostomy (PJ) and end-to-side hepaticojejunostomy were created. An end-to-side DJ was created in the normal fashion. However, due to the patient's previous esophagectomy, the duodenal stump was located very high in the abdomen, just inferior to the crus of the diaphragm, making the DJ anastomosis technically difficult (Fig. 3). A right-sided drain and a left-sided drain were placed; the left drain passed between the PJ and the DJ, which, due to post-esophagectomy anatomy, were in close proximity.

Final pathology of the resected pancreatic mass confirmed poorly differentiated adenocarcinoma with perineural invasion and invasion into peripancreatic adipose tissue (Fig. 3). Genetic analysis was positive for KRAS mutations in codons 12, 13, 61, 117, and 146 of known pathologic significance in pancreatic adenocarcinoma. The pancreatic neck margin on the specimen was positive for adenocarcinoma, but no gross tumor was left in place. The patient recovered uneventfully and per the Whipple accelerated recovery pathway (WARP protocol)⁹ was discharged home on postoperative day 6 on a full liquid diet with his left-sided drain still in place as its fluid was high in triglycerides, suggesting a chyle leak.

The patient was seen in the outpatient office at 2 weeks post-discharge. The drain output was discolored and the patient appeared dehydrated, so he was sent for a CT scan (Fig. 4). This revealed a 7-cm collection with fluid and air near the DJ and PJ. Antibiotics were started. A peripherally inserted central catheter was placed, and total parenteral nutrition (TPN) was commenced. Interventional radiology injected the existing surgical drain with contrast, which opacified the duodenum and gastric pullup, and the drain location was optimized. An UGI swallow revealed extravasation from the DJ, suggestive of a leak. A nasogastric tube was placed into the gastric pullup. A normal gastric pH was achieved by using proton pump inhibitors. Drain output decreased significantly over the course of 3 days. A repeat UGI was performed and revealed a continued leak, with patent efferent and afferent limbs of the anastomosis (Fig. 5). A CT scan revealed a collection remaining posterior to the xyphoid process. The surgical drain was interrogated and left in place. A pigtail drain was placed into the retro-xyphoid collection by interventional radiology.

At this time, the Division of Gastroenterology's Pancreaticobiliary and Advanced Endoscopy section was consulted for endoscopic management. Considerations at that time were primary endoscopic closure, vacuum-assisted closure, or enteral stent placement. The procedure was performed under fluoroscopy. Endoscopy revealed a healthy appearing gastric pullup with patent esophagogastric anastomosis. The defect at the DJ anastomosis was large, and the percutaneously placed pigtail catheter was seen partially within the jejunal lumen. The choice of endoscopic closure was based on anatomic and technical considerations. To this end, the defect was large, at an acute angle to the lumen/axis of the endoscope, and in a small space within which it was difficult to operate. As such, the defect was too large for closure with an over the scope clip (OVESCO) and the space/orientation would not allow for closure by using the endoscopic suturing device (Overstitch, Apollo). Endoscopic vacuum-assisted closure (EVAC) has primarily been described for anastomotic defects closer to the mouth (esophageal) or the anus (rectal), as it requires endoscopically advancing a polyurethane sponge tied to a nasogastric tube into the defect. The sponge must be changed/downsized every 2 to 5 days while the patient remains in the hospital. The duodenal location and difficult access to the DJ anastomosis in this case favored the use of fully covered stents for attempted closure.

A pediatric scope was utilized to navigate the afferent limb, after which a guidewire was passed through the scope and retained. The efferent limb was notably more difficult to navigate but ultimately access was achieved, and a guidewire was advanced for stent placement. A therapeutic upper endoscope was passed over the efferent limb wire and an 18 mm by 103 mm fully covered self-expanding metal stent (FCSEMS; Niti-S, TaeWoong) was placed across the anastomosis into the efferent limb. Similarly, an 18 mm by 100 mm FCSEMS was placed into the afferent limb (Fig. 6). Due to the significant risk of migration of fully covered metal stents, we secured the proximal aspect of the stents to the gastric wall with three sutures by using the endoscopic suturing device (Overstitch, Apollo).

The patient was discharged home a few days after stent placement on continuous TPN with drain care. During his follow-up visit 2 weeks after discharge, his drains had scant output and were removed. He visited a medical oncologist, with plans to discuss adjuvant therapy options for the near future. At the last surgical follow-up on postoperative day 55, he appeared well nourished and his recovery was progressing smoothly. An UGI series with the stents in place revealed no extravasation and good stent location. His TPN was discontinued after 6 weeks, and he has been recommenced on oral alimentation. His stents were successfully removed endoscopically on postoperative day 77, and a post-stent removal UGI series appeared normal with no leaks (Fig. 7).