Does Routine Endoscopy or Contrast Swallow Study After Esophagectomy and Gastric Tube Reconstruction Change Patient Management?

Leakage of the cervical esophagogastrostomy after esophagectomy with gastric tube reconstruction occurs in 5–25 % of patients, and is associated with significant morbidity, and accounts for 25–50 % of postoperative deaths.1‐3 Signs and symptoms of anastomotic leakage are fever, tachycardia, and manifestations at the surgical site including redness, swelling, and drainage of saliva and pus. Appropriate local drainage, intravenous antibiotics, and enteric tube feeding or parental nutrition can manage the majority of anastomotic leakages conservatively. Sometimes, surgical or radiological intervention may be required. In order to detect anastomotic leakage before clinical signs develop and the patients deteriorate, contrast swallow and/or endoscopy are often performed within the first week after surgery. However, it has been reported that contrast swallow studies have a low sensitivity and specificity, failing to contribute to clinical decision making. There is also a risk of aspiration leading to pulmonary complications.4‐8 Upper gastrointestinal (GI) endoscopy has the advantage of direct visualization and quantification of dehiscence, necrosis, or ulcers, and it may be performed in patients who are sedated and intubated.8‐11 On the other hand, there is a fear of iatrogenic injuries and worsening of the anastomotic dehiscence.

The objective of this study is to investigate the diagnostic and predictive value of routine contrast swallow study and endoscopy in the postoperative management of patients with a cervical anastomosis after esophagectomy and gastric tube reconstruction (GTR) for esophageal carcinoma. Our hypothesis was that routine diagnostic studies do not contribute to the early detection of anastomotic leakage.

In this retrospective cohort study, all patients who underwent esophagectomy with gastric tube reconstruction and a cervical anastomosis for esophageal cancer at the Erasmus University Medical Center Rotterdam between January 2005 and December 2009 were included. Patient’s demographics, treatment characteristics were retrieved from a prospective, institutional database. This database includes age, sex, medical history, operative approach, site (neck or thorax) and type of anastomosis (end-to-end or end-to-side), and details on postoperative follow-up including complications and their treatment. As part of the postoperative protocol, patients were scheduled for a contrast swallow study and endoscopy 7 days postoperatively.

Between January 1, 2005 and December 31, 2009, 308 patients underwent esophagectomy with gastric tube reconstruction and a cervical anastomosis. Some 173 patients underwent upper endoscopy, 184 patients underwent a contrast swallow, and 95 patients had both examinations done. The median (range) time between the operation and diagnostic test was 7 6‐12 days. Reasons for delay beyond postoperative day 7 of the contrast swallow or endoscopy were leakage already clinically evident, patient on the ventilator in ICU, logistic reasons, and patients were too sick to undergo an endoscopy or contrast swallow. The logistic regression test performed to compare equality between patient populations in the two groups had a log likelihood of 319,417, no statistical differences were found between groups.

The present study shows that patients without signs or symptoms suggestive of an anastomotic leakage do not benefit from a contrast swallow or upper endoscopy for identifying leaks that require operative or endoscopic interventions. While endoscopy and contrast swallow do show abnormalities in 42 and 20 % of patients, respectively, they do not lead to a change in (conservative) patient management. Hence, aggressive radiological or surgical treatment of patients with abnormal endoscopic findings or contrast swallow does not seem to be indicated. In only a few patients, interventions for anastomotic leakage are needed in due time.

On the other hand, if endoscopy or the contrast swallow study does not show any abnormalities, this does not fully exclude the development of anastomotic leakage and subsequent interventions are needed in 5 and 3 % of patients, respectively (false negative test). The CCI for asymptomatic patients differs significantly for patients without clinical signs of leakage. Even though this difference does not reflect in interventions, it is possible that small, subclinical leakages result in other complications, such as mediastinitis or pneumonia. Furthermore, the CCI is known to be a very sensitive endpoint, as it takes all postoperative complications into account. Still, close surveillance and early recognition of a complication including anastomotic leaks are of utmost importance for best outcomes.

A contrast swallow is the most common routine examination after esophageal surgery. It has several benefits including the low costs and being a relatively safe first-line investigation with a high sensitivity and specificity when interpreted by an experienced radiologist.14 However, the disadvantages of aqueous contrast are that it has a low radiographic density and a low mucosal adherence, thus limiting the ability to detect leaks, particularly in case of subtle ones. Boone et al. presented a low sensitivity and positive predictive value in their series of 207 patients and also reported that 53 % of patients already showed clinical signs of leakage. Doerfer et al. produced comparable results and no longer routinely performed a contrast swallow and preferred a CT with contrast. Tanouchi also reviewed a large series (n = 331) and found a low sensitivity of routine contrast swallow studies.5‐7 Indeed, in the present study, contrast swallow was reported as normal in 3 patients with clinical suspicion of anastomotic leakage, but CT-guided drainage of a mediastinal abscess was needed in 2 patients at a later point in time. In these patients, CT scanning with oral contrast may be superior to a contrast swallow as it allows detection of peri-anastomotic and mediastinal fluid collections that may need surgical or radiological drainage. Endoscopy is likely to be more useful to assess the severity of anastomotic dehiscence in symptomatic patients. In may help in selecting patients that need surgical revision of the anastomosis including resection of an ischemic segment of the gastric tube. These findings are also supported by Oezcelik 10 and Maïsh.11 Schaible 8 and Page et al. 9 presented their retrospective data and concluded that endoscopy is a safe and accurate method to detect early signs of leakage. However, 2 of 8 patients with a normal endoscopy developed a clinical leak that needed a surgical intervention and radiological drainage.

The present study supports the findings of a recent prospective trial that compared the accuracy of contrast swallow, CT with oral contrast and endoscopy for the identification of anastomotic leaks following esophagogastric surgery.15 The authors concluded that routine tests of the anastomotic integrity are unnecessary and, when clinical suspicion is high for a anastomotic leak, CT scan is likely the first modality to perform. Our data that supports also another study has shown that flexible upper gastrointestinal endoscopy is more specific in comparison with a contrast swallow study. While it did not improve the identification of clinical anastomotic leakage, it was beneficial to detect gastric necrosis or ulcers and guide management of these patients.

There are several limitations of the present study. Being of retrospective nature, clinical management of anastomotic leaks may have changed over time. Use of self-expandable stents may have led to a decrease of surgical and radiological interventions for anastomotic dehiscence. However, within the time period of the study, the surgical experience has not changed and also the care pathway has remained the same over time. In order to determine the accuracy of diagnostic tests for the assessment of anastomotic leakage, it is of great importance to define the study endpoint in a consistent and unambiguous matter. In the present study, the definition of Bruce et al. was used. However, given the retrospective design of the study, misclassification cannot be ruled out. The Erasmus Medical Centre is a high-volume specialized center for upper GI surgery and radiologists, and gastroenterologists are well trained in the recognition and treatment of postoperative complications. Hence, our data are likely to be externally valid for other specialized centers.