Feasibility, Effectiveness, and Safety of Endoscopic Vacuum Therapy for Intrathoracic Anastomotic Leakage Following Transthoracic Esophageal Resection: A Prospective Study

Anastomotic leakage (AL) in the upper gastrointestinal (GI) tract is associated with high morbidity and mortality rates. Especially intrathoracic anastomotic leakage leads to life-threatening adverse events. Endoscopic vacuum therapy (EVT) for anastomotic leakage after transthoracic esophageal resection represents a novel concept. However, sound clinical data are still scarce. This prospective, single-center study aimed to evaluate the feasibility, effectiveness, and safety of EVT for intrathoracic anastomotic leakage following abdomino-thoracic esophageal resection.

leakages after esophageal resections. With regards to the upper GI tract, Wedemeyer et al. described, for the rst time in 2008, the endoscopic insertion of a polyurethane sponge to the defect side and the application of an external vacuum (37). The negative pressure therapy decreased and prevented bacterial contamination of the wound and promoted perfusion and granulation of the defect. Hence, the combination of the EVT with defect closure and an effective internal drainage were introduced. Different case reports or case series show favorable results to EVT in terms of sealing rates. Reported treatment success ranges from 66.7-100% (38,39). However, evidence on EVT application in the upper GI tract is still rare and mostly based on small retrospective study cohorts with heterogeneous esophageal defects of different etiologies, e.g. benign perforations and postoperative anastomotic leakages (12,24,25,39).
To date, there is still no consensus on the therapeutic regimen of intrathoracic anastomotic leakages. The aim of this study was to evaluate the feasibility, effectiveness, and safety of EVT in terms of success rate and associated mortalities and morbidities. This prospective single-center study focused on intrathoracic anastomotic leakages following transthoracic esophageal resections.

Patients And Methods
The reporting of this study conforms to the STROBE statement (40).

Study design and population
This single-center study was conducted at the Interdisciplinary Endoscopy Center of the University Hospital of Heidelberg. Data from all patients during the study period from March 2014 to September 2019 were prospectively collected and retrospectively analyzed. The study protocol conformed to the Declaration of Helsinki. It was approved by the Institutional Ethics Committee of the University of Heidelberg (S-635/2013). The trial was registered at the National Library of Medicine. A written informed consent was obtained from each patient or from his legally authorized representative.
Unlike several other retrospective studies on EVT, the inclusion criteria for this homogenous study population were only on intrathoracic anastomotic leakages after transthoracic esophageal resections and subsequent treatment with EVT. Anastomotic leakage was diagnosed either by endoscopy, computed tomography (CT) scan, detection of gastrointestinal content or methylene blue after oral application via drains or by air outlet via drains after air insu ation during EGD. Patients with persistent leakage after prior revisional surgery (n=2) or prior placement of a fully-covered, self-expanding metal stent (SEMS, n=4) ("Niti-S™ Esophageal Stent", TaeWoong Medical, South Korea) were also included in this study. Exclusion criteria included etiologies for esophageal leakages other than postoperative genesis. Examples of these are spontaneous or iatrogenic causes. Initial operations other than transthoracic esophageal resections were also excluded. The patients' demographic and clinical characteristics were retrieved. Furthermore, the leakage characteristics, surgical, and EVT procedural data were collected and analyzed.

Primary and secondary endpoints
The primary endpoint of the study was a successful leakage closure rate with EVT. This was de ned as endoscopically-veri ed resolution of the leakage and the presence of surface epithelium on the former defect. EVT failure was de ned as follows: persistent leakage or stula after termination of EVT; change of treatment strategy; need for surgical reoperation due to anastomotic leakage after EVT; and, death before con rmation of healing.
Secondary endpoints included the feasibility of the EVT procedure (application of the sponge), duration of EVT, number and frequency of sponge changes, procedure-related adverse events of EVT (i.e. endoscopic treatment-associated adverse events such as bleeding and peri-interventional associated adverse events such as aspiration and oxygen desaturation during sedation), duration of intensive care unit (ICU) and intermediate care unit (IMC) hospitalization, overall hospital stay length, course of in ammatory markers during EVT (i.e. white blood cell (WBC) count and levels of c-reactive protein (CRP)) and procedure-related and in-hospital mortality.
Surgical procedure of abdomino-thoracic esophageal resection All patients underwent elective esophageal resection with abdomino-thoracic incision, combined with immediate reconstruction, using a tubularized stomach. The blood supply was based on the right gastroepiploic artery (41,42). A combination of midline laparotomy and right thoracotomy was performed for esophageal resection and two-eld lymphadenectomy. For the intrathoracic anastomosis at the apex of the chest, a circular stapler was used. The stomach was divided with a linear stapler to resect the lesser curvature and the adjacent lymph nodes (43,44). A nasogastric tube was placed into the proximal tubularized stomach and two basal pleural drains were inserted. The postoperative patient management was standardized. From days 1 to 5, only water and tea were allowed. From day 6 onwards, the oral intake was increased in a stepwise fashion.

Endoscopic vacuum therapy (EVT)
Endoscopic vacuum therapy was performed as previously reported and described by Wedemeyer et al (37,45). If the leakage size was large enough and the cavity was accessible to the endoscope, the sponge was inserted into the abscess cavity (intracavitary localization of the sponge). The intracavitary localization of the sponge was documented by CT scan at the beginning of EVT. It was adjusted in case there was close proximity to large vessels. With a diminishing defect size, the sponge was transferred from its initial intracavitary position into an intraluminal position. For cases with only small defects which could not be passed by the endoscope initially, the sponge was placed into the lumen of the esophagus to cover the leak (intraluminal localization of the sponge). The intracavitary-positioned sponge was changed after 3 days. The intraluminal-positioned sponge was changed after 5-7 days. These changes were done until the defect size became too small for further sponge placements and until the defect was lined with surface epithelium (Fig. 1).
The endoscopic procedures were performed either under conscious sedation or general anesthesia, depending on the medical condition of each patient. If pleural drains were extracted at the time of the leakage diagnosis and the sponge was placed intraluminally, the decision for additional external drainage was based on the CT-scan report and the clinical course of the patient.
Generally, in our institution, the further management of anastomotic leakage was based on the size and etiology of the leak, the degree of local contamination, and the severity of the associated systemic response. The decisions for further treatment (i.e. conservative, endoscopic (SEMS, EVT), or operative) were made by a group consisting of a surgeon, surgical endoscopist, and anesthesiologist. In 2014, this special team was established and tasked with decision-making regarding cases of suspected leakage in the early postoperative period after esophageal resections to optimize the clinical management of these patients (46).

Statistical Analysis
Descriptive statistics were calculated for all parameters. The results were expressed as means ± SD (standard deviation) or as median and interquartile ranges for continuous variables. For categorical variables, counts and percentages were used. The student's t-test was used to compare the differences between means and medians. A p ≤ 0.05 was considered statistically signi cant. The analyses were performed using the SPSS software (SPSS Inc., Chicago, IL version 1.07).

Results
A total of 259 patients (217 men, 42 women) with a mean age of 62 years underwent elective esophageal resection through an abdomino-thoracic incision at our institution from March 2014 to September 2019 ( Fig. 2). Seventy-two patients (27.8%) suffered from an anastomotic leakage. Of these 72 patients, eight (11.1%) were treated conservatively. Eleven patients (15.3%) received SEMS therapy. EVT was applied as the primary therapy for anastomotic leakage in 49 patients (68%). Two patients (2.8%) received revisional surgery while another two patients (2.8%) underwent esophageal diversion. A change of therapy was encountered in six patients, who were either initially treated by SEMS (n=4) or had an unsuccessful revisional surgery (n=2) and received EVT later. Six patients underwent esophageal diversion after unsuccessful EVT (n=5) or SEMS therapy (n=1). The overall successful treatment of all anastomotic leakages was achieved in 87.5% of patients (n=63). The overall collective mortality rate was 3.9% (n=10).

Feasibility of Endoscopic vacuum therapy (EVT)
EVT data are shown in Table 2. Technically, EVT could be performed on all patients included in this study.
Effectiveness of EVT: Successful closure rate and failure of therapy The overall successful healing of the anastomotic leakages was achieved in 49 of 55 patients (89.1%).
These patients had a reduction of the cavity size and closure of the leakage by granulation tissue (Table  3). Two patients with EVT as the second-line therapy after unsuccessful revisional surgery and four patients with a therapeutic switch to EVT after failure of the SEMS treatment were treated successfully with EVT.
The EVT overall failure rate of this study was 10.9% (n=6). Five patients (9.1%) received an esophageal diversion: two of them had an necrosis of the conduit (3.6%); another 2 patients (3.6%) showed persistent elevated in ammatory markers during EVT and no further response to EVT during multiple endoscopic interventions; and, the last of the ve patients (1.8%) suffered from symptomatic bleeding during EVT. For this patient, the CT-scan revealed a pseudoaneurysm of the thoracic aorta. It was successfully bridged with a thoracic endovascular aortic repair (TEVAR), followed by an esophageal diversion, and a successful restoration of bowel continuity. The sixth patient died during EVT. This patient suffered from an unclear cardiac arrest. The reanimation was also unsuccessful. The relatives denied autopsy; hence, the underlying cause of death remains unclear.
In ammatory markers during EVT White blood cell count (WBC) and c-reactive protein (CRP) levels are shown in Figure 3. CRP values (Fig.   3A) and WBC count (Fig. 3B) were measured at the beginning of EVT, daily during EVT, and during the rst day following the termination of EVT. All patients showed signs of a systemic in ammatory response at the beginning of EVT with elevated CRP levels frequently accompanied by leukocytosis. The median CRP value at the beginning of EVT was 202.5 mg/l (range, 18-385 mg/l). The median WBC count at the beginning of EVT was also elevated at 14 /nl (range, 6-41 /nl). Both in ammatory markers decreased during EVT treatment. The median CRP value was 30.5 mg/l on the rst day following the termination of EVT (range, 2-98 mg/l) while the median WBC count was 8.5 /nl (range, 4-14.7 /nl). Both in ammatory markers showed a signi cant reduction after the termination of EVT compared to values at the beginning of EVT (CRP: p ≤ 0.001; WBC: p ≤ 0.001).
Safety of EVT: Procedure-related and in-hospital morbidity and mortality As shown in Table 4, the overall EVT-associated adverse event rate in our study was 5.4% (n=3): One adverse event (1.8%) during EVT was the minor bleeding of the above-mentioned patient which led to the diagnosis of an aortoesophageal stula successfully treated by TEVAR and esophageal diversion. This patient recovered fully thereafter and achieved restoration of the bowel continuity by a colon pull-up eight months later. Minor procedural adverse events occurred in 2 other patients (3.6%) during EGD with conscious sedation (oxygen desaturation in one patient and aspiration in the other). Both patients recovered fully with intermittent assisted ventilation via respiratory mask after the termination of endoscopy. They underwent the EVT procedure under general anesthesia thereafter without any further adverse events.
The in-hospital mortality rate was 7.2% (n=4). In contrast to the above-mentioned patient who suffered a cardiac arrest under ongoing EVT, three other patients died after successful EVT and endoscopicallyassured leakage closure. The causes of deaths in these patients were septic shock with multi-organ failure (MOF), urosepsis, and severe acute respiratory distress syndrome (ARDS), respectively.
Recently, there has been growing evidence on the effectiveness of EVT as an initial therapy for anastomotic leakages after esophageal resections. Although in contrast to stent placement, EVT requires multiple endoscopic procedures, there are several advantages of EVT over stent therapy: the vacuum system provides optimal drainage of the wound and an effective sepsis control in case of mediastinitis. Perfusion is also promoted. These culminate in granulation formation and healing of the defect (36). The wound cavity can be visualized regularly and deterioration may be detected early.
The observed successful closure rates and effective sepsis control of EVT were accompanied by a rapid decrease in in ammation markers (CRP and white blood cell count). In accordance with the study results of Laukoetter et al. (12), we showed a signi cant decrease in both in ammatory markers compared to initial values after the termination of EVT. Several studies show higher success rates of endoscopic vacuum therapy compared to stent placement for anastomotic leakages. Berlth et al. included only patients after upper GI surgery for malignancies and compared 77 SEMS-treated patients with 34 EVT-treated patients (47). The success rate of EVT was more favorable at 85.7% compared to SEMS treatment (72.4%). However, this result was not statistically signi cant. This study included the highest number of patients to date. Although only patients with anastomotic leakages after oncological gastroesophageal surgery were analyzed, it included not only patients with intrathoracic anastomosis, but also patients with cervical and abdominal leakages.
A signi cantly higher closure rate of EVT compared to SEMS therapy for treatment of intrathoracic leakages was found by Brangewitz et al. (84.4% vs. 53.8%, respectively) (51). The authors compared 39 stent-treated patients with 32 EVT-treated patients. They also detected a higher stricture rate after SEMS treatment. However, the patients' cohorts of this study were heterogeneous since patients who underwent esophageal resection, gastric fundoplication, iatrogenic perforation, and those who developed Boerhaave syndrome were included in the analyses.
Recently, Rausa et al. performed a meta-analysis of four studies and found signi cantly higher closure rates, shorter treatment durations, and lower adverse event and mortality rates for EVT compared to those for SEMS for esophageal leakages of different etiologies (60).
Our current study only included patients with intrathoracic anastomotic leakages after esophageal resections. We focused only on abdomino-thoracic incisions. In our study population, we showed a successful EVT closure rate of 89.1%. Compared to our own homogenous historic patient cohort (n=32) (article in preparation) about intrathoracic anastomotic leakages following abdomino-thoracic esophageal resection and treatment with SEMS placement, a success rate of 62.5% was detectable. This was notably lower than the success rate of EVT in our study (89.1%). Overall, the in-hospital mortality rate of patients with EVT therapy in our study was 7.2% while the mortality rate in our historic SEMS-treated patient cohort was clearly higher at 15.7%. These data are consistent with the above-mentioned studies. They indicate a favorable effect of EVT compared to stent placement in terms of successful closure, adverse events, and mortality rates for intrathoracic anastomotic leakages (36,51,59).
Although the data from our SEMS group are not novel, these de nitely compare two homogenous patient populations with only intrathoracic anastomotic leakages after abdomino-thoracic esophageal resections. These procedures were performed at our institution according to the same standardized procedures. Although the success rate of 62.5% in our stent cohort was at the lower end of the reported success rates of other groups evaluating only SEMS therapy for esophageal leakages (26,30,31), it is in congruence with studies comparing EVT and SEMS placement (36,51,59). All comparative studies have the same limitations. A historical SEMS-treated patient cohort served as the comparison group. This may have led to a disruption of stent therapy at an earlier stage as soon as EVT was available. Without EVT serving as the alternative therapeutic option, the SEMS treatment would likely have been continued to avoid esophageal diversion.
With regards to the prospectively collected data, but retrospective analysis our cohort study has certain limitations. These include the non-randomized design and limited number of patients included. On the other hand, no prospective randomized or comparative trials for this topic have been published. Our study presents the largest single-center cohort of a homogenous population of EVT-treated patients suffering intrathoracic anastomotic leakage after abdomino-thoracic esophageal resection. Apart from this, considering that the above-mentioned studies showed an apparent superiority of EVT compared to stent therapy, one can doubt how ethical a randomized trial in this context might be.

Conclusion
Endoscopic vacuum therapy is a feasible, effective, and safe method of endoscopic treatment for intrathoracic anastomotic leakages following abdomino-thoracic esophageal resections. This strategy leads to high rates of anastomotic healing and has the potential to reduce the overall mortality. The study protocol conformed to the Declaration of Helsinki. It was approved by the Institutional Ethics Committee of the University of Heidelberg (S-635/2013). The trial was registered at the National Library of Medicine. A written informed consent was obtained from each patient or from his legally authorized representative.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.