Anastomotic Complications After Esophagectomy: Frequency, Prevention, and Management

Classification and Incidence

The incidence of anastomotic leak after esophagectomy ranges from 0% to 35%. Analyses of STS database results reported a 12% incidence of leaks requiring medical or surgical intervention. Variability in the definition of anastomotic leak has led to wide discrepancies in reported rates. Bruce et al. determined that fewer than 40% of articles pertaining to anastomotic leak after esophagectomy included the criteria used to establish the diagnosis, and another systematic review noted definitions included in only 28.3% of studies with 22 different descriptions used.

Possible explanations for differences in reported outcomes in the literature include variable diagnostic tools used to identify a leak, time frame used for evaluation (e.g., 30 vs. 90 day), and anastomotic location. There have been multiple groups that have sought to reach a consensus on a definition for anastomotic leak. The International Multispecialty Anastomotic Leak Global Improvement Exchange group (IMAGINE) defined gastrointestinal postoperative anastomotic leak as “a defect of the integrity in a surgical join between two hollow viscera with communication between the intraluminal and extraluminal compartments.” In 2015 the Esophagectomy Complications Consensus Group (ECCG), a group of 21 high-volume esophageal surgeons from 14 countries, met to create a standardized system for defining and recording complications after esophagectomy ( Table 43.1 ).

Another commonly referenced classification system was proposed by Lerut et al. ( Table 43.2 ). In this definition, clinically occult leaks diagnosed on routine postoperative imaging comprise the most minor end of the spectrum while conduit necrosis is the most severe presentation. The type of leak has significant clinical implications, with grade I/II being relatively low risk and grade III and IV having mortality rates near 60% and 90%, respectively.

Risk Factors

Failed anastomotic healing is affected by multiple interrelated factors. An understanding of these predictive variables is important in the preoperative setting because modifiable factors may be better optimized, and fixed characteristics may be used for risk stratification and patient selection in attempts to minimize poor outcomes.

Esophagectomy is most often used to treat esophageal cancer and also has a role in management of motility disorders; therefore it is not surprising that patients undergoing esophagectomy often present with malnutrition. Poor nutritional status, commonly defined as hypoalbuminemia or weight loss, has previously been demonstrated to be associated with increased risk of anastomotic leak. Analysis of the STS database found that heart failure, hypertension, renal insufficiency, and procedure type were predictive of anastomotic leak. Use of neoadjuvant therapy has been associated with increased risk of leak in some series, although this appears to be dependent on radiation dose given. A review of 1939 patients demonstrated no increase in complications in patients who underwent preoperative neoadjuvant therapy. The presence of diabetes mellitus or advanced patient age surprisingly does not seem to affect leak.

The esophagus has several unique characteristics that make anastomosis after esophageal resection more technically challenging. Unlike the other organs of the gastrointestinal system, the esophagus does not have a serosal layer. This, in combination with the longitudinal orientation of the muscle fibers of the esophagus, contributes to increased fragility and decreased suture security. Attempts to create additional protection using a free peritoneal patch did not impact the rate of leak but did increase the rate of stricture. Furthermore, the esophagus is primarily located within the thorax, which may create a unique challenge for esophageal anastomoses when compared with other types of gastrointestinal leaks. The negative intrathoracic pressure may draw gastric fluid across anastomotic lines and result in leakage. Such leaks into the pleural space may affect the dynamics of intrathoracic pressure and result in respiratory and hemodynamic instability.

The ability of an anastomosis to heal successfully is directly associated with maintenance of adequate tissue perfusion to the conduit. In the course of gastric mobilization and resection of the lesser curvature, the major vessels are ligated; this leaves approximately 60% of the gastric tube to be supplied by the right gastroepiploic vessels and the remaining proximal portion is supplied by small collateral vessels. As such, the anastomosis, which is usually made at the fundus, is within the most ischemic part. Caution must be exercised to minimize trauma to the collateral vessels during dissection. Tension on the anastomosis can lead to impaired blood supply and result in ischemia. Some authors have suggested that the utilization of the whole stomach as opposed to a gastric tube may better preserve blood supply by not disturbing collateral circulation. Arguments against use of the whole stomach claim that use of a gastric tube yields less gastric distention and decreased surface area for acid secretion, both of which can be detrimental to the anastomosis and also lead to poor quality of life. A narrow gastric conduit with longer length can better reach the cervical anastomosis site without tension but must be balanced against excessive narrowing leading to impaired blood supply. Controversy remains as to whether other types of conduit, most commonly colonic, have a similar incidence of leak; however, colonic interposition does require three anastomoses as opposed to one when using a gastric conduit.

Cervical anastomosis has been associated with a higher incidence of leakage than thoracic, with rates ranging from 2% to 26%. This may possibly be explained by the longer distance required for blood supply to travel to the anastomosis site in the neck. In comparison, thoracic anastomoses are associated with leak rates of 0% to 9.3%. There was previously suggestion in the literature that thoracic leaks are associated with higher postoperative mortality, but more recent data do not support this finding when considering early (30-day) outcomes. The higher leak rate associated with cervical anastomosis is more accepted because it is generally contained and thus associated with lower risk. Thoracic anastomotic breakdown is often associated with mediastinal soilage and may have a more severe clinical course.

The type of anastomotic technique selected most often depends upon the surgeon's preference. Stapled anastomotic technique has been used for approximately 25 years and began to increase in popularity due to easier reconstruction, especially in areas where exposure and access are limited. Heitmiller et al. described an 0.8% leak rate after two-layer handsewn anastomosis. Blackmon et al. published a propensity-matched analysis comparing outcomes between side-to-side stapled anastomosis, end-to-end circular stapled anastomosis, and handsewn, with no significant difference in leak rate noted. Multiple randomized control trials have compared various techniques in attempt to identify the safest approach but have failed to identify a significant difference in leak rate. Reconstruction can be performed either via an anterior retrosternal route or a posterior mediastinal route when performing a cervical anastomosis. Urschel et al. performed a meta-analysis that failed to demonstrate a significant difference in leak rate; however, the posterior route is alleged to have the advantages of shorter distance to the anastomosis, resulting in decreased tension, avoidance of foregut angulation, less cardiopulmonary morbidity, soft tissue coverage of potentially devascularized airways, and preservation of the skeletal structures of the thoracic inlet. Anterior reconstruction is primarily justified due to the minimization of tumor recurrence by avoiding the posterior tumor bed.

Laparoscopic surgery has become the standard approach to many gastrointestinal operations because it has been demonstrated to be safe with better postoperative results. Similarly, there has been increased utilization of minimally invasive approaches to esophagectomy. A 2016 Cochrane Review evaluating open versus laparoscopic THE observed fewer overall complications and decreased length of stay. Biere et al. performed a randomized control trial comparing open versus minimally invasive transthoracic esophagectomy and demonstrated significant postoperative benefits, including decreased incidence of pulmonary infections, decreased length of stay, decreased operative blood loos, and improved quality of life at 6 weeks postoperatively and postoperative pain scores. In addition, no differences were demonstrated in number of lymph nodes harvested and rate of R0 resection.

Prevention of Leak

Preoperative patient optimization with correction of malnutrition, optimization of medical conditions, and decreasing steroid use have been demonstrated to help decrease risk of leak. Procedures performed at low-volume centers have been associated with increased risk of postoperative and long-term mortality, as well as increased risk of any (and severe) leaks, which highlights the importance of careful surgical technique in minimizing the incidence of leak. Intraoperatively, there have been attempts to better characterize perfusion to the anastomosis because there is no standardized therapy. Multiple approaches exist, including assessment of graft color, temperature, and checking Doppler signals. Doppler is limited to assessment of macrocirculation. Fluorescence imaging is a promising approach to assessment of the microcirculation and macrocirculation of the gastric conduit. Weaker perfusion, as assessed by intraoperative laser-assisted fluorescent dye angiography, has been demonstrated to be correlated to leak and may be a promising approach.

Following stomach mobilization, a 50% drop in gastric fundus oxygen tension has been measured, and further study has described a direct correlation between intraoperative gastric fundus oximetry and resultant anastomotic success. Ischemic preconditioning has been described as an attempt to preoperatively redistribute gastric blood supply. Embolization of the left and right gastric and splenic arteries was performed 2 to 3 weeks before esophagectomy, thus leaving the stomach dependent on the right gastroepiploic artery. Results in animal models have demonstrated decreased anastomotic leak using these techniques, but similar results have not been demonstrated in humans. An additional technical consideration includes division of the interclavicular ligament to alleviate conduit venous obstruction.

Postoperatively, gastric distention may contribute to anastomotic tension and inhibit venous drainage. Minimization of distention with use of nasogastric tube and promotility agents may help to alleviate this. Pyloromyotomy or pyloroplasty at time of original procedure may also reduce incidence of this complication. In addition, use of pharmacologic agents has been previously studied. Administration of prostaglandin E ₁ within an hour of gastric tube creation has been shown to increase blood flow to the tissue but has failed to show a clinical benefit to date.