Evaluation of Esophageal Anastomotic Integrity With Serial Pleural Amylase Levels.

Background: An anastomotic leak is the most devastating and potentially fatal complication after esophagectomy. Current detection methods can be inaccurate and place patients at risk of other complications. Analysis of pleural fluid for amylase may be more accurate and place patients at less of a risk for evaluating the integrity of an esophageal anastomosis.

Methods: We retrospectively reviewed prospective data of 45 consecutive patients who underwent an Ivor Lewis esophagectomy over an 18-month period and evaluated their anastomotic integrity with serial pleural amylase levels (PAL).

Results: There were 40 men (89%), and median age was 63 years (range, 35 to 79). Indication for esophagectomy was cancer in 38 patients (84%); 27 (71%) underwent neoadjuvant chemoradiation. A barium swallow was performed in the first 25 patients at median postoperative day (POD) 5 (range, 5 to 10); the swallow was negative in 23 patients (93%). Serial PALs were obtained starting on POD 3 and stopped 1 day after toleration of clear liquids. The PALs in the no-leak patients were highest on POD 3 (median 42 IU/L; range, 20 to 102 IU/L) and decreased (median 15 IU/L; range, 8 to 34 IU/L) to the lowest levels 1 day after clear liquid toleration (p = 0.04). Two patients had a leak and had peak PALs of 227 IU/L and 630 IU/L, respectively; both leaks occurred on POD 4, 1 day before their scheduled swallow test. The last 20 patients underwent serial PALs only, without a planned swallow test or computed tomography scan for anastomotic integrity evaluation. One of these patients had a leak on POD 5 with a low PAL of 55 IU/L the day before the spike of more than 4,000 IU/L. Two of the leaks were treated with esophageal stent placement and intravenous antibiotics, and the remaining patient's leak resolved with intravenous antibiotics, no oral intake, and observation only. None of the leak patients required transthoracic esophageal repair or drainage of an empyema. There was 1 postoperative death (2%) secondary to aspiration pneumonia on POD 10; no leak was ever identified, and the patient had been eating for 3 days before death. Complications occurred in 15 patients (33%), most commonly respiratory; no respiratory issues occurred in PAL-only evaluated patients. No late anastomotic leaks occurred in any patient while in the hospital or after discharge.

Conclusions: Serial PALs for the detection of esophageal anastomotic leaks proved to be accurate, safe, and inexpensive. Elimination of barium swallows and computed tomography scans for evaluation of anastomotic integrity may decrease aspiration risks as well as associated pulmonary failure during the postoperative period. Serial PALs may be the preferred method of detecting an anastomotic leak after esophagectomy. A prospective randomized study is warranted.