Between 2010 and 2020, all patients who underwent Ivor Lewis esophagectomy for an underlying malignancy were included in this study. All patients who developed post-surgical anastomotic leaks were further analyzed. This led to a total of 28 patients who were treated for anastomotic leak at our hospital. In addition, the following parameters were examined: date of surgery, postoperative day of detection of an anastomotic leak, CAES classification, type of treatment used, median combined intensive care unit and intermediate care stay, median hospital stay, tumor histology, tumor grading, neoadjuvant therapy, number of lymph nodes harvested, R-status, operation method, operation time, morbidity, and mortality. Regarding endoscopic treatment, the following variables were analyzed: location of the defect, size of the defect, number of stents or vacuum sponges used, event-related complications, length of treatment, and treatment outcome.

An anastomotic leak was defined as a communication between the intra- and extraluminal compartments through a defect in the integrity of the intestinal wall of the anastomosis. Routine examination using a dynamic swallow study was performed until 2016. If the dynamic swallowing study suggested the presence of an anastomotic leak, it was followed by an upper endoscopy (UE) or computed tomography (CT). After 2016, routine examination of the anastomosis was abolished and was only performed if patients showed symptoms suggestive of a leak. If so, a combination of UE and CT was performed. Subsequently, patients with macroscopic visible mediastinal leakage cavity (referred to as the “extraluminal cavity”) were always treated using EVT. In contrast to this group, patients with smaller anastomotic defects and none or small leakage cavity (called “intraluminal cavity”) were treated with stent therapy. In addition, EVT has become increasingly established as a standard therapy over the past few years.

If an anastomotic leakage was clinically suspected or confirmed using a CT scan, EVT was evaluated as a therapeutic option. A sufficient external thoracic drainage was inserted in the case of a huge mediastinal septic abscess. In cases of confirmed or suspected anastomotic leakage, UE was performed in sedated or mostly intubated patients (GIF-H180, GIF-H190; Olympus Co., Tokyo, Japan). If there was evidence of a large extraluminal leakage cavity, which could only be inspected using a small-caliber nasal endoscope (GIF-N180) but not with a normal endoscope (GIF-H180, GIF-H190), the defect was expanded using balloon dilatation (CRE™ Wireguided 12–15 mm; Boston Scientific, Marlborough, MA, USA). During the initial endoscopy, the leakage cavity was cleaned and measured to determine the required length and diameter of the sponge, which was then reshaped accordingly. Open-pore polyurethane sponges, Eso-SPONGE^® (B. Braun Melsungen AG, Melsungen, Germany) with a primary diameter of 24×55 mm and a 12 CH Redon drain or an individually adapted sponge (V.A.C. VERAFLO™ Dressing Kit; KCI, St. Paul, USA) fixed to a drain (Argyle™ Edlich Gastric Lavage Tube; 16 CH, Medsitis, USA), were used.

In general, the intraluminal placement of the sponge in the case of small anastomotic defects (usually less than 8 to 10 mm) or residual cavities with no infection can be differentiated from the intracavitary placement of the sponge, where it is introduced through the wall defect into the extraluminal, i.e., mediastinal, cavity. The intracavitary version of EVT was preferred. For placement of the sponge, two endoscopic methods were used, the “push” technique or the “piggyback” technique. Using the push technique, the sponge was advanced to the correct location along an overtube with a pusher or the endoscope, and a specially approved device (Eso-SPONGE^®; B. Braun Melsungen AG, Melsungen, Germany). Using the piggyback technique, the sponge was placed in the leakage cavity under direct endoscopic vision, while a suture loop placed at the tip of the sponge was grasped using endoscopic forceps and the sponge was pulled close to the endoscope. While the first technique is often used for small anastomotic leaks with intraluminal positioning of the sponge, the second technique is preferably used for intracavitary placement of the sponge. The drainage tube was placed transnasally and connected to a variable-speed medical vacuum pump (V.A.C. ULTA^®; KCI, San Antonio, Texas, USA). Suction was applied at a negative pressure of 75–125 mmHg. In addition, a transnasal gastric or duodenal tube was inserted for enteral nutrition. After a dwell time of 3–5 days, the next endoscopy was performed. In this procedure, the sponge was removed orally after it was disconnected from the vacuum pump. The treated cavity was then examined using an endoscope to document the success of the treatment, particularly with a focus on subsequent granulation. A new sponge was inserted after re-measuring the size of the cavity to determine the size of the new sponge. To promote effective cavity closure, the diameter of the sponge was first reduced without reducing its length to allow for closure of the remaining channel in subsequent treatment cycles from the distal part of the channel to the proximal part. EVT was continued until the cavity was reduced to less than 1 cm. During each endoscopy, the CAES grading of anastomotic insufficiency in the esophagus was reevaluated retrospectively.

Figure 1 illustrates the typical clinical course of treatment in one of the study patients undergoing successful EVT.

Similar to the principles of EVT, endoscopic evaluation was carried out with regard to the size of the defect, existence of an extraluminal leak cavity, and perfusion of the anastomosis or the gastric sleeve. SEMS were mostly inserted under direct endoscopic view in the intensive care unit (ICU) without radiological control. First, a stiff wire (Amplatz Super Stiff Guidewire; Boston Scientific) was placed down to the stomach under endoscopic control. A fully or partially covered SEMS with a diameter of 22–28 mm was inserted and released under endoscopic view (Wall Flex™ Esophageal Stent, partially covered, 22–28 mm, Boston Scientific) with a total length of 100 mm. The SEMS was removed after approximately three weeks. In the case of persistent insufficiency, another stent was inserted. CAES grading was reevaluated retrospectively.

Figure 2 illustrates the management of an anastomotic leak in one of the study patients using an SEMS.

Statistical analysis was performed using IBM SPSS Statistics Version 24 64-Bit-Version for Mac OS (IBM Co., Armonk, NY, USA). Continuous variables are presented as medians. To compare these variables, we employed analysis of variance (ANOVA) with multiple factors. Categorical variables were compared using the chi-squared test. Statistical significance was defined as p<0.05.

Between 2010 and 2020, 141 patients underwent Ivor Lewis esophagectomy for underlying malignancy. All relevant patient characteristics are presented in Table 1. A total of 28 patients were diagnosed with postoperative anastomotic leakage, resulting in an anastomotic insufficiency rate of 19.9%. Of these patients, three patients were treated conservatively, 13 patients were treated with EVT, seven patients were treated with SEMS, one patient was treated with a hemoclip, and four patients received primary surgery to treat the defect (Fig. 3). Six patients required surgical revision after the initiation of endoscopic treatment. All 28 patients were classified using the CAES classification.

Of the 28 patients, 23 were men and five were women. The median age was 58.5 years (range: 32–75 years). The median body mass index and American Society of Anesthesiologists classification were 25 kg/m² and 2, respectively. The reasons for esophagectomy were adenocarcinoma in 25 patients (89.3%) and squamous cell carcinoma in three cases (10.7%). Twenty of the 28 patients received neoadjuvant chemotherapy or radiotherapy. All but one patient underwent open Ivor Lewis esophagectomy with a median operation time of 290 min (range: 144–624 min).

Of the 141 patients, 28 (19.9%) were diagnosed with postoperative anastomotic insufficiency. The median time from surgery to diagnosis was 7.5 days (range: 2–30 days). The median distance from the upper incisor to the defect was 25 cm (range: 18–30 cm), with a median defect size of 10 mm (range: 5–30 mm). In 10 patients, the defect developed into a macroscopic visible extraluminal cavity. The median hospital stay was 48.5 days (range: 9–193 days) with a median ICU/intermediate care (IMC) stay of 22 days (range: 9–193 days). Of the 28 patients who developed anastomotic insufficiency, 20 (71.4%) were treated successfully, while eight (28.6%) patients died. The overall endoscopic findings are presented in Table 2. Eight patients were treated with alternative methods to EVT or SEMS. Of these eight patients, four underwent primary surgery, two of whom required additional surgical revision and died postoperatively. Three of the eight patients were treated conservatively, and one patient was treated with endoscopic clipping of the defect. All patients receiving non-surgical alternative treatments were successfully treated. The median combined IMC/ICU and hospital stay were 12 and 29 days, respectively. The median defect size was 3.5 mm (range: 2–4 mm) for the conservative group and 22.5 mm (range: 10–25 mm) for the surgical group.

Thirteen patients (46.4%) with a mediastinal leakage cavity (extraluminal cavity) were treated with EVT. The median time to diagnosis of anastomotic leak after the primary surgery was 8 days. All patients with an extraluminal cavity were treated using EVT, and all of those within the EVT group who died had an extraluminal cavity. No EVT-related complications were observed. The median time for EVT was 24.5 days (range: 8–80 days) with a median of five exchanged sponges (range: 4–18). The median defect size was 15 mm, and the median defect was located 26 cm from the upper incisor. Eight patients were successfully treated (61.5%), while five patients passed away. However, complete closure of the defect was achieved in 12 patients (92.3%). Five patients needed surgical revision during EVT, and 4 of those patients died. All of these patients required surgery due to complications that could not be attributed to EVT. The reasons for surgery were ischemia of the colon, necrosis of the pancreas, postoperative incarcerated hiatal hernia, and hemothorax. The median ICU/IMC stay was 38 days (range: 9–193 days), with a median hospital stay of 74 days (range: 9–193 days). The overall outcomes of EVT are shown in Table 3.

Seven patients (25%) were treated with SEMS. In the stent group, none of the patients had an extraluminal cavity. The median duration of SEMS therapy was 22 days (range: 3–31 days), with a median of one SEMS exchange (range: 1–2). The median defect size was 6 mm, and the median defect was located 23 cm from the upper incisor. Six patients were successfully treated (85.7%), while one patient required surgical revision and died. The median ICU/IMC stay was 20 days (range, 16–57 days) with a median hospital stay of 41 days (range: 22–123 days). The median time to diagnosis of anastomotic leak was 7 days. During SEMS treatment, event-related complications, including stent migration (n=1) and perforation (n=1), were noted. A close defect was achieved in six out of seven patients (85.7%). The overall outcomes of SEMS therapy are shown in Table 3.

Statistical analysis to compare SEMS and EVT revealed that there was a statistical significance (p<0.05) in terms of defect size, presence of an intraluminal or extraluminal cavity, and sponges/stents used. No other difference in outcomes between the two treatment options were statistically significant. The results of the statistical analysis comparing the two modalities are presented in Table 3.