Classification and Management of Leaks after Gastric Bypass for Patients with Morbid Obesity: A Prospective Study of 60 Patients

Transcription

1 OBES SURG (2012) 22: DOI /s CLINICAL REPORT Classification and Management of Leaks after Gastric Bypass for Patients with Morbid Obesity: A Prospective Study of 60 Patients Attila Csendes & Ana Maria Burgos & Italo Braghetto Published online: 23 March 2012 # Springer Science+Business Media, LLC 2011 Abstract The most important and frequent major complication after gastric bypass is the appearance of a leak, which can result in death of a patient. The purpose of this prospective study was to determine the incidence of a postoperative leak, to propose a classification and to evaluate the results of conservative or surgical treatment. All patients submitted to gastric bypass either laparotomic or laparoscopic were included in a prospective protocol. In all radiological evaluation at 4 th day after surgery was performed. The presence of a leak was evaluated according to the day of appearance, its location and its severity. Results of medical or surgical treatment were analyzed. From 1764 patients submitted to Roux-en-Y gastric bypass, 60 had a postoperative leak (3.4%). This leak appeared early after surgery (before 4 days) in 20%. It was a localized subclinical leak in 20% and clinical septic in 80%. There were 7 possible anatomic location of a leak, being the gastrojejunal anastomosis the most frequent location (53%) followed by gastric pouch. The highest mortality was associated to the jejuno-jejunal anastomosis. Conservative treatment was employed in near 65% of the patients: The mean time of closure of a leak was 34 days. The appearance of a postoperative leak is a major and serious complication. It can be classified according to the day of appearance, its severity and its location. Conservative or surgical treatment can be employed properly if these 3 parameters are carefully evaluated. A. Csendes (*) : A. M. Burgos : I. Braghetto Department of Surgery, Hospital J.J. Aguirre, University Hospital, Santos Dumont 999, Santiago, Chile acsendes@redclinicauchile.cl Roux-en-Y gastric bypass, performed either by laparotomic or laparoscopic approach, is widely performed for patients with morbid obesity and is considered as the gold standard operation for these patients [1 4]. This procedure is a complex operation, especially when performed by laparoscopic approach. The most important and frequent major surgical complication is the appearance of a leak after surgery, which can result in death of the patient [5 7]. Therefore, the purpose of the present prospective study was: (a) to establish the incidence of this complication after laparotomic and laparoscopic approach, (b) to propose a classification of these leaks based on day of appearance, severity, and location of the leak, and (c) to evaluate the results of medical or surgical treatment of this complication. Material and Methods Patients Studied The present paper corresponds to a prospective protocol, which was established on August 1999 and includes all patients operated on until December They corresponded to 1,764 patients submitted to Roux-en-Y gastric bypass either open or laparoscopic. All patients had a BMI 40 or >35 Kg/m 2 plus co-morbidities such as diabetes, hypertension, or dislipidemia. The exclusion criteria were the presence of alcoholism, drug addiction, or psychological disorders. Preoperative endoscopy was performed in all patients, according to our protocol of preoperative work in patients with morbid obesity candidates to surgical treatment [8, 9]. If Helicobacter pylori was present, it was treated with antibiotics for 10 days in all patients in whom the excluded stomach was left in situ.

2 856 OBES SURG (2012) 22: Surgical Procedure There were 1,191 patients submitted to laparotomic approach and 573 patients to laparoscopic approach. Among the patients with an open procedure, 962 were submitted to resectional gastric bypass [8] and 229 patients to standard gastric bypass. Among the 573 patients with laparoscopic approach, 386 corresponded to classic gastric bypass and 187 to resectional gastric bypass. The complete details of the open procedure have been described previously in detail [6, 8]. Basically, a small gastric pouch (20 30 ml) was constructed. The distal gastric segment was removed, dividing the duodenum with a GIA-60-mm stapler (Covidien, Mansfield, MA). The gastrojejunostomy was performed using a circular stapler No. 25 (Covidien), leaving an internal diameter of 15 mm. The staple line was reinforced with continuous absorbable sutures of Byosin 4 0 (Covidien). The length of the Roux limb was 125 to 150 cm. The same procedure was performed by laparoscopic approach, employing the technique described by Almino Ramos et al. [10]. Basically, the small gastric pouch is constructed employing endo GIA stapler (Covidien). The gastrojejunostomy is made with an antecolic jejunal limb 60 to 70 cm distal to the Treitz angle with lineal white stapler No. 30. The hole made by the stapler is closed either with Byosin or with Vycril 3 0. Then, the jejunojejunostomy is performed by lineal white stapler 45, 125 to 150 cm distal to the gastrojejunostomy in a similar way. With this technique, both anastomoses can be checked with the administration of 60 ml of methylene blue injected through a nasogastric tube by the anesthesiologist, in order to confirm the impermeability of all suture lines. Then, the afferent loop is transected with a white stapler 60, creating in this way the same Roux-en-Y gastric bypass as in open procedure. One or two silastic drains were left at the side of the gastrojejunal anastomosis. Radiological Evaluation In all patients, a radiological study was performed at the 4th day after surgery with liquid Barium sulfate. With the patient standing, 20 ml of barium is swallowed, and three characteristics of the small gastric pouch were evaluated: size, emptying, and impermeability of the pouch. A total of six different pictures are taken in left lateral, right lateral, posterior, and supine positions. Two hours after, a plain X- ray of the abdomen is taken in order to evaluate the distal jejunojejunal anastomosis and the progression of the barium until the colon. We have demonstrated previously that a small, localized leak can be easily diagnosed with barium sulfate, but not with liquid contrast medium such as Gastrographin or Hypaque [11]. Classification of Leaks The presence of a leak was evaluated by three parameters [6, 7, 11, 12]: 1. Time of appearance after surgery, dividing them in early (1 4 days after surgery), intermediate (5 to 9 days after surgery), and late (10 or more days after surgery) Fig. 1 Identification of each possible site of a leak after gastric bypass, with or without gastric resection, according to oro-caudal direction. Type I: gastric pouch. Type II: gastric jejunal anastomosis. Type III: jejunal stump. Type IV: jejunojejunal anastomosis. Type V: excluded stomach (inexistent in gastric resection). Type VI: duodenal stump (inexistent in nonresectional bypass). Type VII: blind end biliary jejunal limb

3 OBES SURG (2012) 22: Table 1 Clinical features and postoperative mortality among 1,764 patients with morbid obesity submitted to gastric bypass No leaks Postop leaks p Value N=1704 N=60 Gender Women 1,329 (78%) 40 (67%) n.s. Men 375 (22%) 20 (33%) Mean age 40.8± ±11.5 n.s. Preop BMI (kg/m 2 ) (25.4%) 19 (31.7%) ,038 (60.9%) 34 (56.7%) (13.6%) 7 (11.6%) n.s. Postop mortality 0 6 (10%) n.s. not significant 2. Severity of the leak in types I or II [8]. Type I corresponds to a small localized leak, with none or minimal systemic repercussion, producing either a localized fluid collection managed by percutaneous drains or easily managed by the surgical drain left in situ. Type II corresponds to a leak producing systemic repercussion, with fluid and air collections which needs aggressive surgical or medical treatment with enteral or parenteral nutrition, antibiotics, and careful management of drains. 3. Location of the leak: our classic approach was to deal with anastomotic leaks. However, the careful observations of the precise location of a leak after Roux-en-Y gastric bypass has revealed that eventually there are seven potential sites for a leak, as shown in Fig. 1, described from proximal to distal: Type 1 Gastric pouch Type 2 Gastrojejunal anastomosis Type 3 Jejunal stump Type 4 Jejuno jejunal anastomosis Type 5 Excluded stomach Type 6 Duodenal stump (in resectional bypass) Type 7 Blind end biliary jejunal limb after laparoscopic surgery. Eventually, there may be two leaks present. Surgical Management of Leaks When surgical treatment was necessary, it consisted in open or laparoscopic reoperation, identification and suture of leaks, lavage with large amount of saline, placement of nasojejunal enteral feeding tube, use of methylene blue to check impermeability of suture lines, and placement of two drains around the site of the leak. Conservative Treatment After the demonstration of the presence of a leak, a nasojejunal tube was placed either under radiological control o guided by endoscopy. No oral feeding was established, and enteral feeding was started in a similar protocol than patients submitted to total gastrectomy for gastric cancer [13]. The drains were left in situ until less than 20 ml of fluid in 24 h was collected [14]. If no drain was present, by percutaneous approach, a 12-F drain was inserted inside the liquid collection. If fever was present and bacteria were isolated, antibiotics were employed according to their sensitivity. The evolution and closure of the leak was evaluated every 14 days by radiological methods (abdominal scanner and barium sulfate administration). Statistical Analysis For statistical significance, the Fisher test and the Chi square test were employed, considering a p<0.05 as significant. Table 2 Clinical and laboratory characteristics of patients with leaks after gastric bypass N=60 Symptoms and signs Mean value±s.d. Number of patients Mean day appearance after surgery Fever (>37.5 C) 38.4±6 ( ) 49 (81.7%) 9.3±10.5 (1 57) Abdominal pain 40 (66.7%) 8.3±12.5 (1 56) Tachycardia >100 bpm 118±12.4 ( ) 45 (75.0%) 10.9±13.8 (1 58) Leucocytosis >10,000/mm ±4450 ( ) 43 (71.7%) 11.6±12 (1 51) C-reative protein >11 mg/l 257±124 (83 549) 60 (100%) 13±1 (1 58) Anemia due to hemoperitoneum 4 (6.7%) 1.5±1 (1 3) S.D. standard derivation

4 858 OBES SURG (2012) 22: Table 3 Diagnosis of leaks (n=60) Methods Number of patients Percent Mean day after surgery Radiology with barium sulfate Abdominal scanner ,5 Increase daily output through drain Appearance of biliary aspect Appearance of methylene blue Appearance of food through drain Results Table 1 shows the main clinical features of 1,764 patients submitted to Roux-en-Y gastric bypass, separated in the group of no leaks and the group of leaks. There were 60 patients (3.4%) with postoperative leaks, with similar age, gender, and BMI distribution compared to no leaks. There were six postoperative deaths (0.3%), all of them due to the appearance of leaks. There was no difference in the incidence of leaks in patients with BMI over 50 kg/m 2,or associated co-morbidities such as hypertension or type 2 diabetes. The abuse of nicotine was not evaluated. Table 2 demonstrates the main clinical and laboratory characteristics. The most frequent symptom was fever (over 37.5 C), starting as early as the 1st postoperative day. Early tachycardia (over 100 bpm) with absence of fever was another important parameter to think in the possibility of a leak. C-reactive protein was elevated in 100% of the patients, although it increases normally up to 5 days after surgery in non-complicated patients [15]. There were four patients who developed acute postoperative anemia due to hemoperitoneum, all of them reoperated 1 to 3 days after surgery. In Table 3, we demonstrate the early and initial method for diagnosis of a leak. According to our protocol of performing upper radiology in all patients at the 4th postoperative period or earlier if there is suspicion, in all 17 patients with a leak appearing before the 4th day after surgery, radiology was 100% correct. However, if a leak appeared late than the 4th postoperative day, obviously the radiological evaluation was normal at that time. None of the patients had barium peritonitis. This fact was very important in order to establish the occurrence of leaks later than the 5th postoperative day. Therefore, in these patients, Table 4 Classification of leaks according to day of appearance after surgery N=60 Type leak Number of patients Percent Early (1 4 P.O.D) Intermediate (5 9 P.O.D.) Late (10 or more P.O.D.) P.O.D. postoperative day abdominal scanner was very useful, demonstrating the presence of air and fluid collection around the surgical site. In seven patients, the increase of the daily output through the drain suggested the presence of a leak [14]. In a similar way, the change of a serohematic aspect of the fluid to bilious aspect confirmed the presence of a leak, mainly from the duodenal stump. In three patients with symptoms, the oral administration of methylene blue and the immediate appearance of it through the abdominal drain confirmed the presence of a leak. Finally, in three patients (5%), the appearance of food through the drain clearly demonstrated a leak. The classification of the leaks according to the day of appearance after surgery is shown in Table 4. Early leaks corresponded to 28%, in a similar proportion to late leaks. Intermediate leaks were the most frequent type. The precise day of appearance of a postoperative leak is seen in Table 5. This table demonstrates that leaks can appear as late as 59 days after surgery. Table 6 demonstrates the classifica- Table 5 Time of appearance of leak (postoperative day) N=60 Postoperative day Number of patients

5 OBES SURG (2012) 22: Table 6 Classification of leaks according to severity N=60 Type leak Number of patients Percent Mortality I (subclimical localized) II (clinical septic) (12.5) tion according to the severity or systemic repercussion of a leak. There were 20% of type I localized leaks, while the usual more disseminated and clinically relevant leaks were 80%. Table 7 shows a very important detail which is the precise location of a leak and its mortality. As shown in Fig. 1, there are seven possible locations of a postoperative leak after gastric bypass, with different mortality rate. The most frequent site of a leak is the gastrojejunal anastomosis (53%) and the second is at the gastric pouch. However, there are other sites of location of a leak, and the most serious is a jejuno-jejunal leak with a high mortality rate. There were 11 duodenal stump leaks among the 1,149 patients submitted to resectional gastric bypass (0.9%). Double leaks occurred in three patients. Table 8 shows the daily output of abdominal drain in 359 patients with uneventful evolution [14] compared with the 60 patients with leaks. It can be seen that at 5th day after surgery, the drain is retired with a daily output of 20 ml. On the contrary, in patients with leaks, this daily amount of fluid collection was increased. Table 9 demonstrates the treatment of patients according to the day of appearance of the leak and its severity. It can be seen that among patients with type I leak, only one patient was reoperated on early after surgery, while the other 11 patients were managed conservatively. On the contrary, among patients with type II leak, 42% were reoperated on, but 58% were managed conservatively. Table 10 shows the duration of the leaks, that is, the time elapsed between the diagnosis and closure of the leaks. The mean closure time was 34 days. Discussion The results of the present study suggest, in first place, that at least in our experience, the occurrence of a postoperative leak is a major complication after gastric bypass, requiring close clinical observation, great surgical care, and patience, because the healing or the closure takes time. In second place, the diagnosis of a leak can be made by several methods, together with careful clinical evaluations. In third place, we propose a classification of postoperative leaks based on three parameters: time of occurrence, clinical severity, and precise location. In fourth place, based in this classification, not all patients require reoperation but rather an aggressive medical conservative management. Finally, we have measured the precise day of closure of a leak, which has not been determined in a precise way before. The review of several surgical studies [6] has shown that the incidence of postoperative leaks after Roux-en-Y gastric bypass is variable, from 0% to 5.6% with a mean of 2.4%. It represents a major and serious surgical complication and in our experience; it was the cause of death in all six patients who died after surgery. Among the 1,764 patients submitted to gastric bypass, other complications such as pulmonary embolism have occurred but were not mortal. Why such a different statistics report from 0% up to 5.6% of leaks? It is difficult to know, but we wonder how leaks were diagnosed, how precise the methods were, and which was the clinical observation. Several papers report postoperative fluid collections and abscesses, which in our experience always have corresponded to minimal leaks with inflammatory response producing such collections. It is hard to believe 0% of leaks. This never happens in Table 7 Classification of leak according to location and mortality N=60 Location leak Number of patients Mortality Type I gastric pouch 11 (18.3%) 0 Type 2 gastrojejunal anastomosis 32 (53.3%) 3 (9.3%) Type 3 jejunal stump 1 (1.7%) 0 Type 4 jejuno-jejunal anastomosis 3 (5.0%) 1 (33.3%) Type 5 Excluded stomach 1 (1.7%) 0 Type 6 Duodenal stump 9 (15.0%) 2 (22.2%) Type 7 Blind end jejunum 0 Type 8 double leaks 3 (5.0%) 0 Types 2 and 6 2 Types 4 and 6 1

6 860 OBES SURG (2012) 22: Table 8 Daily output of abdominal drain after gastric bypass (ml/ 24 h) Uneventful postoperative evolution N=359 P.O.D. 1st day nd day rd day day day day Retired 129 7th day 74 8th day th day th day day day day day day 97 Postoperative leak N=60 surgery because postoperative complications are inherent to a complex surgery. Therefore, we come to the problem of an early and precise diagnosis of the occurrence of a leak. Usually, all of us employ radiological methods such as contrast studies and CT scan. However, the early and initial suspicion must be made on clinical parameters such as fever the 2nd and 3rd day without other explanation, upper abdominal pain without peritoneal signs and tachycardia, in absence of other clinical parameters. In these patients, we perform Table 9 Treatment of leak type I and II according to day of appearance N=60 Classification Type I N=12 Type II N=48 Early (1 4 P.O.D.) N=17 Reop 1 9 Conservative 3 4 Intermediate (5 9 P.O.D.) N=28 Reop 0 6 Conservative 7 15 Late ( 10 P.O.D.) N=15 Reop 0 5 Conservative 1 9 Total reop 1 (9%) 20 (41.6%) Conservative Reop. reoperation, P.O.D. postoperative day immediately a CT scan and afterwards a contrast study using Barium sulfate and not other liquid contrast. Why is this? Because we have clearly seen that small localized leaks are not diagnosed by liquid contrast medium but clearly shown with Barium sulfate, similar to what was reported before, with 17% to 22% of positivity employing Gastrographin [16, 17]. Besides, it is important to detail that none of our patients with leak had the so called barium peritonitis. As the leak was early and diagnosed promptly with barium sulfate, a great proportion with patients with type II leaks was reoperated on, performing an aggressive cleaning and lavage with saline solution. In the majority, a small portion of barium was left in situ, which appeared in consecutive radiological evaluations, being encapsulated and without any complication. Other simple clinical observations are also very useful for establishing the diagnosis of a leak, such as the increase of daily output through the drains, appearance of biliary aspect or food through the drains, or the easy oral administration of methylene blue and its immediate appearance in the abdominal drain. We have measured the daily output of the serohematic fluid through a right and left abdominal drain in 359 patients submitted to gastric bypass who had a normal postoperative course [14]. The mean collected fluid corresponded to 50 to 90 ml/day the first day after surgery, while in the 4th day, it was 20 ml/24 h, moment at which they were retired. In the present study, we have proposed three parameters to classify the behavior of a postoperative leak. In first place, we have seen a different behavior and different treatment whether a leak is early, intermediate, or late after surgery. Not all leaks appear the first 4 days after surgery, as has been reported in many surgical reports. On the contrary, only 28% are early and the great majority occur later, even when patients are ready to leave the hospital or appearing at home. The later a leak occurs, the easier the Table 10 Duration of leaks (time between diagnosis and closure of leaks) N=54 (excluded death) Duration (day) Number Percent Mean 34.4±37 (range 7 160)

7 OBES SURG (2012) 22: treatment and the better the prognosis. Therefore, it can be understood easily that the radiological control with liquid contrast performed the 1st day after surgery, according to the protocol of several surgeons, is useless because on the 1st day after surgery, only four out of 60 leaks occurred; therefore, this study will miss the great majority of leaks. Perhaps the only use of this early control is to check the emptying of the small pouch. In second place, it is important to precise which is the magnitude or clinical severity of a leak. In this particular aspect, we have employed our classification of leaks after total gastrectomy for gastric cancer [11], a clinical situation similar to a gastric bypass, in which only 2% to 5% of the entire stomach is employed as a small gastric pouch. The type I leak corresponded to a small, localized leak, without septic or severe clinical manifestations, and the type II leak, which may have diffusion to abdominal or pleural cavity with severe clinical and septic manifestations. However, the most important contribution is the identification of the precise site or location of a leak after surgery. The majority of surgical publications, which mention postoperative leaks, do not precise their location and simply mention leak as a single complication. However, according to our classification, there are seven potential sites for the appearance of a postoperative leak, with different clinical evolution, management, and mortality. A leak at the small gastric pouch, at the jejuno-jejunal anastomosis, or at the duodenal stump is completely different when performing gastric resection. We have mentioned the importance of the location of a leak in previous publications [6, 7]. Therefore, all these eventual points must be clearly checked by radiological contrast medium such a barium sulfate or by CT scan. We have demonstrated in the present study that contrary to what has been postulated, a great majority of patients with postoperative leaks can be managed conservatively, as we have shown previously [6, 7] and do not require surgical re-intervention. In proposing this management, it is very important to consider the day of appearance, the severity, and the location of a leak. If it is early after surgery, it is type II and is located at the jejuno-jejunal anastomosis, or at the gastrojejunal anastomosis, prompt surgical repair is needed. On the contrary, if a leak appears several days after surgery, even if it is a type II, usually located at the gastric pouch or gastrojejunal anastomosis, it can be managed with percutaneous guided placement of a drain, enteral feeding through nasojejunal tube and close clinical observations. None of the patients managed conservatively has died, similar to what we reported in 16 patients with leaks after sleeve gastrectomy [12]. We have not employed an expandable stent in any of our patients, which could be another conservative alternative management of a localized leak. It is important to detail that the incidence of leaks did not depend on the retrocolic or antecolic approach. In all patients submitted to open gastric bypass, a retrocolic route was employed, according to our large previous experience, in the surgical treatment of 1,076 peptic ulcer patients [18] or in 210 patients with long segment Barrett s esophagus [19]. On the contrary, in all patients submitted to laparoscopic approach, the antecolic route was employed, demonstrating similar incidence of leaks compared to the retrocolic route. Finally, it is interesting to determine the closure time of a leak appearing after gastric bypass, which is a mean of 34 days after the diagnosis, being shorter than the mean healing time after sleeve gastrectomy, which is 45 days [12]. The close review of several articles has shown that there is no information regarding this specific point. We believe that it is important to know it, because the psychological alterations which develop in the postoperative period after the occurrence of a leak may be severe in the patients and in the close family. The knowledge of the healing time could be important for all subjects involved in the complicated management of a patient with postoperative leak. Conflicts of interest of interest. References The authors declare that they have no conflict 1. Brolin RE. Gastric bypass. Surg Clin N Am. 2001;81: MacLean LD, Rhode BM, Nahr CW. Late outcome of isolated gastric bypass. Ann Surg. 2000;231: Fobi MAL, Lee H, Holness R, et al. Gastric bypass operation for obesity. World J Surg. 1998;22: Buchwald H, Williams SE. Bariatric surgery worldwide Obes Surg. 2004;14: Baker RS, Foate J, Kemmeter P, et al. The science of stapling and leaks. Obes Surg. 2001;14: Csendes A, Burdiles P, Burgos AM. Conservative management of anastomotic leaks after 557 open gastric bypasses. Obes Surg. 2005;15: Csendes A. Conservative management of anastomotic leaks. Obes Surg. 2006;16: Csendes A, Burdiles P, Papapietro K, et al. Results of gastric bypass plus resection of the distal excluded gastric segment in patients with morbid obesity. J Gastroint Surg. 2005;9: Csendes A, Burgos AM, Smok G, et al. Endoscopic and histologic findings of the foregut in 426 patients with morbid obesity. Obes Surg. 2007;17: Ramos AC, Galvão Neto MP, de Souza YM, et al. Laparoscopic Duodenal Jejunal Exclusion in the Treatment of Type 2 Diabetes Mellitus in Patients with BMI<30 kg/m 2 (LBMI). Obes Surg. 2009;19: Csendes A, Díaz JC, Burdiles P, et al. Classification and treatment of anastomotic leaks after extended total gastrectomy for gastric carcinoma. Hepatogastroenterology. 1990;37:174 7.

8 862 OBES SURG (2012) 22: Csendes A, Braghetto I, Leon P, et al. Management of leaks after laparoscopic sleeve gastrectomy in patients with obesity. J Gastrointest Surg. 2010;14: Papapietro K, Díaz E, Csendes A, et al. Early enteral feeding in patients with total gastrectomy for gastric cancer. Rev Méd Chile. 2002;130: Csendes A, Burgos AM, Burdiles P, et al. Use of abdominal drains after gastric bypass for morbid obesity. Rev Chil Cir. 2006;58: Csendes A, Burgos AM, Roizblatt D, et al. Inflammatory response measured by body temperature, C-reactive protein and white blood cell count 1,3 and 5 days after laparotomic or laparoscopic gastric bypass surgery. Obes Surg. 2009;19: Gonzalez R, Nelson LG, Gallagher SF, et al. Anastomotic leaks after laparoscopic gastric bypass. Obes Surg. 2004;14: Hamilton EC, Siurs TL, Hamilton TT. Clinical prediction of leak after laparoscopic Roux-en-Y gastric bypass for morbid obesity. Surg Endosc. 2003;17: Csendes A, Braghetto I, Burdiles P, et al. Morbidity of elective Surgery for duodenal ulcer ( ). Rev Chil Cir. 1995;47: Csendes A, Burdiles P, Braghetto I, et al. Early and late results of the acid suppression and duodenal diversion operation in patients with Barrett s esophagus: Analysis of 210 cases. World J Surg. 2004;26: