Diagnosis and Management of Anastomotic Leaks after GB

Abstract & Commentary

By Namir Katkhouda, MD, FACS, Professor of Surgery, Chief, Minimally Invasive Surgery, University of Southern California (USC)-Los Angeles, CA. Dr. Katkhouda is a consultant for Baxter, Ethicon, Storz, and Gore.

Synopsis: Lack of specificity in clinical presentation and imaging studies make diagnosing anastomotic leaks challenging, so operative exploration should be part of the diagnostic algorithm.

Source: Gonzalez R, et al. Diagnosis and contemporary management of anastomotic leaks after gastric bypass for obesity. J Am Coll Surg. 2007;204:47-55.

Background: Anastomotic leaks are a serious complication of bariatric surgery. The objective of this study was to describe the clinical presentation and outcomes of treatment in patients who develop anastomotic leaks after Roux-en-Y gastric bypass.

Methods: In this study, 3018 consecutive patients, from prospective databases, who underwent Roux-en-Y gastric bypass in 4 tertiary referral centers, were reviewed.

Results: Sixty-three patients (2.1%) developed anastomotic leaks (open, 2.1%; laparoscopic, 2.1%) at a median of 3 days (range 0 to 28 days) after Roux-en-Y gastric bypass. Symptoms and signs included tachycardia (72%), fever (63%), and abdominal pain (54%). Upper gastrointestinal series and CT demonstrated leaks in only 17 of 56 (30%) and 28 of 50 (56%) patients, respectively; when done jointly, both studies were negative in 30% of patients. The 68 anastomotic leaks occurred at the gastrojejunostomy (49%), excluded stomach (25%), jejunojejunostomy (13%), gastric pouch (9%), and uncertain location (4%). Forty patients (63%) required 58 reoperations for drainage of intraabdominal collections (55%), repair of anastomotic defects (34%), or revision of the leaking anastomosis (11%), with an overall morbidity of 53% and an overall mortality of 10%. Nonoperative treatment was successful in 23 of 26 patients, with no mortality and an overall morbidity of 61% (P = NS versus operative). Operative treatment was more common in patients with hypotension or oliguria (P < 0.01).

Conclusions: Lack of specificity in clinical presentation and imaging studies make diagnosing anastomotic leaks challenging, so operative exploration should be part of the diagnostic algorithm. Non-operative treatment is safe and effective in a subset of patients who are known to have controlled leaks which exhibit stable hemodynamic parameters.


This study first shows that the incidence of leaks is around 2%; it is identical for open or laparoscopic surgery. The number for laparoscopy is low, and the equivalent of the open technique reflects that the operations were performed by skilled laparoscopic surgeons. Most leaks occurred on the third day, but the range shows that some can occur up to 28 days after surgery, though it is rare.

It is interesting to note that 49% of the leaks occurred at the gastrojejunostomy site, even though the anastomosis was checked intraoperatively every time using insufflation with an air underwater seal. This leads me to believe that this technique is not accurate enough to detect anastomotic imperfections. I prefer the use of methylene blue, which has the advantage of giving indication of the volume required to fill the pouch. Postoperatively, everybody agrees, and this study is no exception; the most reliable sign of a leak is Tachycardia, as defined by a pulse above 100.

One has to bear in mind that this has to be weighed by the patient's perioperative pulse rate. I have had many patients with a pulse of 105 that alarmed us for a leak, although all other vitals were normal, and some studies that were done were negative. The patients were discharged the next day without any problem. Nevertheless, a pulse of 120 postop is ominous and should prompt additional studies, unless the patient's condition warrants an immediate re-exploration (low urine output, hypotension, surgical abdomen). In all other patients, an upper GI is ordered but, in my experience, confirmed by the current study, it is of poor yield for leaks. I would suggest skipping this test, which will only falsely reassure the surgeon in most cases, and order a CT scan with oral and IV contrast.

I agree with Gonzalez and colleagues that an attempt for conservative, non-operative management can be attempted if the patient's vitals are stable. In their hands, non-operative management was successful in 23 of 26 patients without mortality. If no drain is left in during surgery, a percutaneous one placed under CT can save the day.

Now, a word about the white count. Many studies emphasize the elevated white count and the left shift. I would like to insist, and this was not mentioned in the study, that a patient can present postoperatively with a normal to low white count. Very little is mentioned about the possibility of bandemia, which would close the case in favor of an intraabdominal catastrophe.

In the event of a reoperation, Gonzalez et al insist on the placement of a gastrostomy tube for decompression, as well as early feeding through the remnant.

The closure of a gastrointestinal defect at the anastomoses will not only make matters worse, it's generally ineffective because of the amount of inflammation. A good drainage, with an omental patch, is a good option.