Marshall JS, Srivastava A, Gupta SK, Rossi TR, DeBord JR. Roux-en-Y Gastric Bypass Leak Complications. Arch Surg. 2003;138(5):520-524. doi:10.1001/archsurg.138.5.520
Copyright 2003 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2003
Enteric leakage is a significant complication of the Roux-en-Y gastric bypass (RYGB) procedure that can be treated successfully.
Retrospective study of 400 consecutive RYGB patients from 1999-2002.
Community hospital with a university surgical residency.
Hospital records of 400 morbidly obese patients who underwent gastric bypass surgery were reviewed.
Main Outcome Measures
Time of discovery of leak, location of leak, treatment, hospital stay, and mortality.
Twenty-one patients (5.25%) developed leaks. The mean body mass index (calculated as weight in kilograms divided by the square of height in meters) was 54.2. Thirteen patients were noted to develop a leak at the gastrojejunal anastomosis, with an average time to diagnosis of 7.0 days. Five of these patients underwent reexploration, and 8 were successfully treated with percutaneous drainage alone. Four patients developed leaks at the jejunojejunal anastomosis (mean time to diagnosis, 2.0 days). All of these patients required exploration, and 2 patients died. Four patients were noted to have leaks in other areas (average time to diagnosis, 3.5 days). Two patients were treated with drainage, and 2 underwent exploration. The average hospital stay of all patients was 33 days.
Enteric leakage is a significant complication of the RYGB. Patients who are suspected of having an enteric leak because of signs of sepsis or hemodynamic instability require emergent exploration. Leaks that are more insidious may be treated successfully with percutaneous drainage. Aggressive exploration of patients who appear to be septic, and percutaneous drainage of insidiously developing leaks may decrease patients' morbidity and mortality.
OBESITY IS reaching epidemic proportions in the United States and throughout the developed world. Dietary therapy is of little help for many people with morbid obesity. In 1967, Mason and Ito1 described the technique of gastric bypass for control of morbid obesity. The Roux-en-Y gastric bypass (RYGB) refined this procedure, which resulted in adverse effects, such as the development of marginal ulceration, bile reflux gastritis, and the risk for a leak when tension was placed on the mesentery of a loop of the jejunum.2 Alden3 described the use of a stapler to partition and not divide the stomach in an attempt to decrease the risk of leaks from a divided gastric pouch. In 1994, Wittgrove et al4 described the laparoscopic approach to the RYGB. The National Institutes of Health (Bethesda, Md) have concluded that bariatric surgery, including the RYGB, is an appropriate form of therapy for morbid obesity in patients who have failed to respond to medical treatment.5 However, surgical therapy can be associated with complications. We present our experience with the complication of anastomotic leak after RYGB.
A retrospective study of 400 consecutive morbidly obese patients who were treated with RYGB was undertaken. These procedures were performed by 4 surgeons between July 1999 and March 2002 in a community hospital with a university surgical residency program. Hospital records were reviewed for patient demographics, body mass index (BMI [calculated as weight in kilograms divided by the square of height in meters]), type of bypass (open vs laparoscopic), primary RYGB vs revision (either conversion from a vertical banded gastroplasty or revision of a prior RYGB), location of the leak, the postoperative time that the leak was discovered, treatment rendered, length of hospital stay, and outcomes.
An open RYGB was performed by all 4 surgeons, and laparoscopic RYGB was performed by 3 of the surgeons. Open procedures were performed with a TA-90B stapler (US Surgical Corp, Norwalk, Conn) and a 2-layer hand-sewn gastrojejunostomy. The efferent Roux limb was 100 cm early in the series but was later consistently measured to 200 cm. The Roux limb was reconstructed in a retrocolic fashion in all patients. Most patients had a retrogastric limb with an anastomosis to the lesser curvature, although several patients had an antegastric Roux limb with an anastomosis to the lesser curvature of the stomach. The jejunojejunostomy was created with a GIA (gastrointestinal anastomosis) -60 linear stapler (US Surgical Corp), and the resulting enterotomy was closed with a TA-55 stapler (US Surgical Corp). Gastrostomy tubes to the distal gastric pouch were used early in the series (42 patients). A closed-suction drain was placed across the gastrojejunal anastomosis at the discretion of the surgeon. No protocol was used to determine the need for drain placement. Drains were usually placed in patients in whom the surgeon felt were at increased risk of leak because of the complexity of the operation or in patients in whom the detection of a leak with radiographic studies would be difficult due to the patient's weight, usually 158.8 kg. Wounds were closed with absorbable internal fascial retention sutures and a continuous running absorbable suture.
Laparoscopic RYGB was performed with 5 operative ports (3 5-mm ports and 2 15-mm ports) placed in the upper abdomen and left anterior axillary line. A separate port was placed in the right anterior axillary line to retract the liver. In the early stages of our series, a 100-cm efferent Roux limb was used in patients with a BMI of less than 45, and a 200-cm limb was used in those with a BMI of 45 to 50. This was eventually standardized to a limb length of 200 cm. The jejunojejunostomy was created using an endoscopic GIA-45 stapler (US Surgical Corp) and a running suture closure of the resulting enterotomy. A retrocolic, retrogastric RGBP was then performed. The gastric pouch was created by serial application of an endoscopic GIA reinforced with Peristrips (Biovascular Inc) around the anvil of a 21-mm end-to-end anastomosis (EEA) stapler, which had been placed via a distal gastrotomy. The EEA stapler was used to create the gastrojejunostomy. This anastomosis was reinforced with sutures, and the mesenteric defects were closed with sutures. A closed suction drain was placed across the gastrojejunal anastomosis. The 15-mm ports were closed with absorbable sutures. The Fisher exact test was used to determine statistical significance between groups.
Four hundred patients (340 women, 60 men; mean age, 43 years [range, 16-69 years]) underwent RYGB for morbid obesity between July 1999 and March 2002 (Table 1 and Table 2). Three hundred twenty-six patients underwent open RYGB and 74 patients underwent laparoscopic RYGB. Three hundred eighty-six were primary procedures and 14 were revision procedures. The mean BMI was 53.2 (range, 35-102). Twenty-one patients (6 men, 15 women) developed leaks (5.25%). The mean BMI of those who had leaks was 54.2. Eighteen patients who had leaks had undergone open procedures (5.5%), and 13 of these were primary procedures (4.2%). Five (35.7%) of the 14 patients who underwent revision procedures developed leaks. Three (4.1%) of the 5 patients who had leaks had undergone primary laparoscopic RYGB. The laparoscopic RYGB patients who developed leaks were among the first 10 patients in our laparoscopic series. Using the Fisher exact test, there was a statistical difference between the rates of leakage in revision vs primary procedures (P<.01).
Thirteen (62%) of the 21 leaks were noted at the gastrojejunal anastomosis. The average time to diagnosis was 7 days (range, 2-14 days). Signs and symptoms included fever (>38.6°C), chills, tachycardia (>110 beats/min), nausea, malaise, shortness of breath (secondary to a reactive pleural effusion), and a change in the nature of the drain effluent. An upper gastrointestinal (GI) series using water soluble contrast followed by barium if no leak was seen were used to confirm leaks in 8 of the 13 patients. Two patients were operated on based on clinical findings alone, and 2 underwent computed tomographic scans and 1 patient underwent an ultrasound as the initial diagnostic test. Five of these 13 patients underwent exploration. The decision to operate was based on the clinical condition of the patient. Patients who were hemodynamically stable, with minimal temperature elevations, were treated with percutaneous drainage if possible. However, patients who had hemodynamic instability or decreased urine output were operated on emergently. At exploration, the leaks were controlled with revision of the anastomosis and placement of closed suction drainage. Eight patients were successfully treated nonoperatively with percutaneous drainage of the leak site and any associated abscesses. Two of these 8 were treated with continued closed-suction drainage via drains prophylactically placed at the time of the RYGB.
Four patients (19%) of 21 were noted to have developed leaks at the jejunojejunal anastomosis. The average time to diagnosis was 2.0 days (range, 1-3 days). These patients exhibited more dramatic early postoperative distress, increased fluid requirements, decreased urine output, and vascular collapse. Diagnostic studies (upper GI and computed tomography) were performed in 2 of the patients, but these tests did not aid in the diagnosis of a leak or the decision to operate. All of the patients required reexploration, and 2 of these patients died of peritonitis secondary to the leak.
Four patients (19%) of 21 were noted to have developed leaks at other sites. The mean time to diagnosis was 3.5 days (range, 2-5 days). Two patients had a leak at the defunctionalized distal portion of the stomach. One of these was demonstrated only at laparotomy and one, radiographically. One patient was noted to have a leak near the gastrojejunostomy, but the exact site of the leak was not identified, even by contrast radiologic studies. The fourth patient was noted at exploration to have a leak in the Roux limb, distant from the jejunojejunostomy. The 2 patients in whom leaks were not explored were successfully treated with drainage. A prophylactic drain was used in one patient and a percutaneous drain was used in the other.
It should be noted that the 2 patients who exhibited symptoms at 13 and 14 days did not have a prophylactic drain placed. One of these was an outpatient who had been feeling ill for several days before coming to the clinic (on day 13), at which time the leak was demonstrated radiographically. The patient who came to the clinic on day 14 had a gastrocutaneous fistula, which might have been found earlier had a drain been left in place at the time of the original operation. The patient who came to the clinic on day 10 had a prophylactic drain in place and noticed a change in the nature of his drainage on day 9, before his clinic visit. Our practice has been to leave prophylactic drains in place for 10 days. No patient has experienced a leak after removal of a prophylactic drain.
Table 3 presents a comparison of the time of discovery of leaks with the location of the leaks and the treatment rendered. Leaks that appeared earlier in the postoperative course tended to be treated surgically, while leaks that were discovered later in the postoperative period were often managed via percutaneous drainage techniques when possible. The leaks at the jejunojejunostomy appeared earlier in the postoperative course and were all treated surgically. This was because of the dramatic clinical findings in these patients. The leaks at the gastrojejunostomy tended to occur later in the postoperative period and were often amenable to percutaneous drainage. The average hospital stay for patients with a leak was 33 days. The average stay of those who did not have leaks was 4.1 days. Analysis using the Fisher exact test reveals that a longer hospital stay due to a leak is statistically significant (P<.01).
Gastrointestinal anastomotic leak after an RYGB is a known complication. The reported incidence is between 1% and 5.1%.6,7 Revision of previous bariatric procedures carries a higher risk of leakage, reported to be as high as 19%.8 We noted a slightly greater than 5% incidence in our overall series. Leaks were noted to be more frequent in the early part of our laparoscopic RYGB experience, which is consistent with that of DeMaria et al7 and with revision bariatric surgery.
In this review, we have identified 2 main areas of leakage, the gastrojejunal anastomosis and the jejunojejunal anastomosis. It has been our experience that distal enteroenterostomy leaks are more rapidly clinically evident (at 2.0 days vs 7 days for gastrojejunal leaks) and are often only diagnosed clinically and confirmed at laparotomy for sepsis, peritonitis, and vascular collapse. We agree with other authors that these clinical signs of a leak should mandate early exploration and control of the leak.6 Radiographic absence of a leak at the gastrojejunal anastomosis does not preclude a distal anastomotic leak.
Gastrojejunal anastomotic leaks may manifest in the perioperative period with signs and symptoms of sepsis or peritonitis, including tachycardia and increased fluid requirements, in which case they require emergent operative intervention. However, 8 of 13 patients developed leaks that were more insidious. Signs and symptoms included fever, malaise, left-sided abdominal pain, and shortness of breath with pleural effusion. These were successfully treated with percutaneous drainage techniques, without the need for surgical intervention. Percutaneous drainage is very advantageous in the control of leaks after RYGB. However, the patient must be clinically stable, without hypotension or oliguria, for this method to be chosen over exploration. We have also explored patients who were clinically stable but were too obese for percutaneous drainage. Two patients developed leaks that were controlled by closed-suction drains that were placed prophylactically at the time of surgery. These leaks were discovered by noting a change in the nature of the drainage; the patients were otherwise asymptomatic. These patients were treated on an outpatient basis and recovered without sequelae.
Fourteen of our patients underwent revision procedures, of whom 5 (35.7%) developed leaks. This increased rate of leakage was statistically significant when compared with our primary procedures. This can be due to the increased dissection required by reoperative surgery, with a resulting increased risk of injury and ischemia to the tissues.
We also noted that the leaks that occurred in our laparoscopic patients occurred early in the series. This may have been due to a learning curve effect. Our laparoscopic patients do not require revisions, and generally have a lower BMI, allowing a better view of the operative field.
Prophylactic drains detected and controlled anastomotic leaks in 2 of our patients, with minimal morbidity to the patients. For this reason, we have begun to use a single closed-suction drain placed across the gastrojejunal anastomosis in all of our reoperative cases and in cases judged by the surgeon to be at high risk for leakage. We also use these types of drains in the extremely obese patient who, because of weight limitations, would be difficult to study radiographically. We had 2 patients with leaks detected after postoperative day 10. These patients did not have prophylactic drains but showed signs that leakage would have been detected earlier had a drain been placed. No patient had a leak develop after a drain was removed at 10 days. We therefore recommend that prophylactic drains be left in place for 10 days.
Enteric anastomotic leakage is a significant complication of the RYGB. Hemodynamically unstable patients who are suspected of having an enteric leak require emergent exploration. Leaks that cause less dramatic symptoms may be treated successfully with percutaneous drainage. Aggressive exploration of unstable patients and percutaneous drainage of stable patients may decrease patients' morbidity and mortality. Prophylactic drains may control leaks without further intervention and should be considered in reoperative surgery or in patients determined by their surgeons to be at risk for developing a leak.
Corresponding author and reprints: J. Stephen Marshall, MD, 420 NE Glen Oak, Suite 301, Peoria, IL 61603 (e-mail: firstname.lastname@example.org).
Accepted for publication January 25, 2003.
This study was presented at the 110th Scientific Session of the Western Surgical Association, Vancouver, British Columbia, November 19, 2002, and is published after peer review and revision. The discussions that follow this article are based on the originally submitted manuscript and not the revised manuscript.
Thomas A. Stellato, MD, Cleveland, Ohio: I would like to emphasize a couple of important points which they presented in their paper. The first finding: the overall leak rate was 5.25%. For those undergoing a revisional procedure, leaks were nearly 7 times as high at 36%. In this latter situation, the authors have recommended routine drains and this probably does have merit.
The second finding was that 30% of their initial 10 laparoscopic procedures leaked. The third finding was that most leaks occurred at the gastrojejunostomy, as would be expected. The authors found that radiology could be helpful, both for diagnosis and for treatment, but not all of these patients required reoperation. Morbidity is significant with length of stay greater than 1 month, but no mortality was noted in this group. In contrast, a small number of patients developed leaks at the enteroenterostomy. These patients became sick quickly. Radiology was generally of little help, and 40% of these patients died. This leads to one of the most important messages of the paper. This is a message that is recognized by surgeons operating on the morbidly obese but still deserves restating. Any patient who is ill after Roux-en-Y gastric bypass with unexplained tachycardia warrants an exploration, even in the face of a normal swallow study. I have the following questions for the authors: (1) Do you test the gastric anastomosis intraoperatively? We use both air and methylene blue to test our open gastrojejunostomies upon completion. (2) Do you perform a swallow study routinely in the postoperative period? We study all of our patients on postoperative day 1 and that has allowed us to discharge 51% of our open gastric bypass patients on postoperative day 2. (3) Were there any differences in the outcome between patients who leaked and patients who did not leak? Specifically, can you discuss stenosis and weight loss? (4) Did you change your technique in the laparoscopic group, especially since 3 of the first 10 patients leaked.
Finally, a complication that you really did not address but I wonder if you could comment upon and that is pouch to gastric fistula. You leave your stomach in continuity as we do also, and I was wondering whether you could tell us a little bit about your incidence of pouch to gastric fistula.
Donald E. Fry, MD, Albuquerque, NM: We have had our experiences with leaks in these patients, even though we do disconnect the pouch from the gastric remnant. Our concerns that Dr Stellato has raised about breakdown of the staple line over time, especially in selected patients where learning new eating habits can be very problematic and will result in some staple line breakdown and potential failures.
I am curious as to whether the authors have reexplored patients based on clinical findings and found nothing. We have had that experience with using clinical criteria. I have had the clinical impression that the very obese patient has a robust septic response, and whether it is a little atelectasis or whatever, I have been impressed that some of these patients on the first and second postoperative evening seem to look extraordinarily bad and we have reexplored them on clinical criteria only to find nothing. So I am curious whether there has been reoperation that found nothing in the patients that appeared to have a robust septic response. There is a lot of interest in the literature right now on inflammation in the obese patient and so there is some basic science evidence that would seemingly indicate that the morbidly obese patient may in fact have a robust inflammatory response to even a fairly minor provocation, and I am curious what the authors' observations are on that.
Finally, the figure of a 5% operative risk continues to be bantered around for this operation. They have presented 2 deaths out of 400 related to leaks. I am curious whether they have had any medical deaths. I have had the unfortunate experience of a patient on the fourth postoperative morning getting his things ready to go home and having a fatal myocardial infarction. We did a patient with a BMI of 117 earlier this year who ended up having the incidental and clinically occult finding of hepatitis C–associated cirrhosis, and she did not fare well either. These patients bring an array of medical comorbidities to the table and I am curious whether the authors would share with us whether there were nontechnical mortalities in the series. I really have the feeling that the current published literature on this operation is understating the aggregate mortality rate that we see in caring for these very high-risk, complicated patients.
Philip E. Donahue, MD, Chicago, Ill: We are going to learn a lot by looking at this series, especially since they describe a spectrum of modes of presentation with the groups of leaks and a way of sorting them out.
Please tell us more about your adjunctive measures to decrease leaks at staple lines. You mentioned one special tape that you use. Do you ever use fibrin glue or should we just use staples and sutures? Along the line of preventing associated morbidities, do you use beta-blockers preoperatively in all of the fat patients, only for the super obese, or do you individualize in all cases? Finally, you haven't mentioned the gallbladder. Does routine cholecystectomy still have a role?
Michael B. Farnell, MD, Rochester, Minn: Of the 400 patients you presented, 7 were approached laparoscopically. What is your preference now? How do you decide whether you are going to perform laparoscopic or open Roux-en-Y gastric bypass? And a corollary question. Have you noticed a decrease in the abdominal wall complications in patients who had the laparoscopic Roux-en-Y gastric bypass? For example, hernias, wound infections, seromas?
Dr Marshall: To answer your questions regarding the testing of the gastrojejunal anastomosis, we did do this early in our laparoscopic series. We did it for a short while in our open series, and did not find that it affected our leak rate. We also felt that since we drained all of our laproscopic patients, after about 50 to 60 laparoscopic procedures without leaks, we chose not to test and haven't found an increase in the leaks.
We do not get a contrast swallow routinely postoperatively. We do get a contrast swallow if the patient has tachycardia or has increased fluid requirements with mild oliguria. If they have severe oliguria and massive fluid requirements, we reexplore them.
In terms of our outcome for the leak vs nonleak patients, our weight loss has been about the same in both groups. With respect to our stricture formation, we didn't look at that specifically but it doesn't appear to have been any higher. We made one small change in our technique in the laparoscopic procedure from the early part of our series after about the first 5 to 10 patients. We paid more attention to the suture reinforcement of the EEA gastrojejunal stapled anastomosis.
We have not looked specifically at our pouch-gastric fistula rate. We have found 2 patients of 400 who have had that who presented with mild weight loss that we weren't happy with, and we studied those patients, so that would be about 0.5% that we found, although we haven't studied a set group of patients to specifically address this.
Dr Fry, we have had at least 1 negative exploration based on clinical findings the day after surgery. We have investigated many patients based on the clinical findings with an upper GI series, so yes, we have found that there are a lot of patients who present with significant tachycardia and mild oliguria, that based on an upper GI study do not have leak. I do agree with your hypothesis that there is a robust inflammatory response in these morbidly obese patients that we do not yet fully understand.
We have had other medical deaths. Our series now is about 550 patients, and our overall mortality rate is 1.35%. We have had patients die from ARDS (adult respiratory distress syndrome), from pneumonia, and we had 1 patient die of liver failure. We have had 1 patient die of pulmonary embolus.
Dr Donahue, the adjunctive means we have used for staple line reinforcement are the Peristrip enhancement of the staple line for the divided gastric pouch and the application of sutures to reinforce the proximal gastrojejunal stapled anastomosis in our laparoscopic cases. That does take tension off the staple line when we add the sutures to the staple line. We do not routinely use beta blockade in these patients unless their cardiologist recommends that we do. We perform cholecystectomy in these patients if they have cholelithiasis preoperatively. We do get an ultrasound in all of our patients. We do not do a cholecystectomy if they do not have cholelithiasis.
Dr Farnell, we have chosen to limit our laparoscopic approach at this point to patients with a BMI of 50 or just slightly above 50. If the patient has had previous upper abdominal surgery, the patients are done open. All revisional surgery is done open. Any patient who has not had previous upper gastrointestinal surgery and who has a body mass index of less than 50 to 52 is considered for laparoscopic gastric bypass. We have noted a definite decrease in the incidence of abdominal wall complications in these patients. One of the most prominent advantages of doing this procedure laparoscopically is the decreased risk of wound infection and subsequent herniation.