Profile of a randomized, controlled trial.
Ijichi M, Takayama T, Toyoda H, Sano K, Kubota K, Makuuchi M. Randomized Trial of the Usefulness of a Bile Leakage Test During Hepatic Resection. Arch Surg. 2000;135(12):1395–1400. doi:10.1001/archsurg.135.12.1395
An intraoperative bile leakage test will assist in decreasing postoperative bile leakage in patients undergoing hepatic resection.
Randomized controlled trial.
One hundred three consecutive patients who were scheduled for hepatic resection without biliary reconstruction. Associated cirrhosis was present in 49 patients (48%), and only 24 (23%) had normal livers. Patients were randomized to undergo (n = 51) or to not undergo (n = 52) a bile leakage test, according to age, liver function, and hepatectomy method. The 2 groups were similar in baseline demographics.
A bile leakage test was carried out by injecting isotonic sodium chloride solution through the cystic duct, and interrupted suturing was taken for a bile leak on the transected liver surface.
Main Outcome Measures
The incidence of postoperative bile leakage and the length of the postoperative hospital stay. Bile leakage was defined as continuous drainage, with a bilirubin level of 86 µmol/L or more (≥5 mg/dL), beyond 1 week.
Twenty-one patients (41%) in the test group showed a bile leak, and a median of 1 site (range, 1-6 sites) was closed during the test. Postoperative bile leakage was observed in 3 patients (6%) in the test group and in 2 (4%) in the control group (P = .99). The odds ratio of the event was 1.53 (95% confidence interval, 0.25-9.44). The median postoperative hospital stay lasted 17 (range, 13-47) and 18 (range, 12-41) days for the test and control groups, respectively (P = .98).
This randomized trial suggested no advantage in using a bile leakage test during hepatic resection.
ALTHOUGH recent progress in surgical techniques and perioperative management has reduced the morbidity and mortality associated with liver surgery,1,2 bile leakage is still one of the most common complications leading to a prolonged hospital stay. The rate of bile leakage after hepatic resection has been reported to be 6% to 11%2- 6 and, to our knowledge, there are no standardized methods for its prevention.
The principal causes of bile leakage are intraoperative biliary injury, insufficiency of the bilioenteric anastomosis, and bile oozing from the transected liver surface. Regarding hepatic resection, the third factor appears to be mostly responsible. An intraoperative bile leakage test has been used to detect leakage from the transected liver surface in several institutions. With this technique, the leak points on the transected liver surface are predicted by injecting solution material through the cystic duct and will be closed steadily beforehand. This approach has been recommended as a useful preventive measure against postoperative bile leakage.1,2,5 However, to our knowledge, there are no prospective studies to confirm the efficacy.
We, therefore, conducted a randomized, controlled trial to assess whether a bile leakage test would help in reducing the incidence of posthepatectomy complications.
The study included any patient with a liver tumor, either primary or metastatic, who was scheduled for elective hepatic resection. Exclusion criteria were as follows: (1) a need for biliary reconstruction, (2) previous surgery of the upper abdomen, (3) concomitant resection of other organs, and (4) emergency operations. Informed consent was obtained from the patients, and this study was approved by the ethics committee.
Randomization was individually performed soon after the surgical indications and method were determined intraoperatively. Patients were randomly allocated to the bile leakage test group or to the control group according to a minimization procedure, with age (<60 or ≥60 years), the indocyanine green retention rate at 15 minutes (<20% or ≥20%), the number of tumors (1 or ≥2), and the hepatectomy method (limited resection and subsegmentectomy or segmentectomy and lobectomy) as stratification variables.
Parenchymatous division of the liver was accomplished using Péan forceps (clamp fracture method) under the intermittent occlusion of the hepatic inflow.7 The principle of the hepatectomy method was an anatomical resection using ultrasonography. Subsegmentectomy was conducted according to the ultrasonically guided staining of the tumor-bearing portal area.8 For segmentectomy or lobectomy, the clear demarcation obtained by hilar dissection or each major hepatic vein exposed on the transection plane was the landmark for anatomical resection.
After parenchymatous division had been completed, any bleeding from the raw surface was stopped and any visible sites of bile leak were closed by interrupted suturing with an atraumatic needle. The bile leakage test was then carried out in the patients assigned to the test group. The catheter with a hole in the proximal side of the balloon (model 0191406, Phycon I.O.C Catheter; Fuji Systems Corporation, Tokyo, Japan) was used.
After cholecystectomy, the balloon catheter was inserted through the cystic duct into the common bile duct. The balloon was inflated, and isotonic sodium chloride solution was slowly injected until it filled the biliary tree, as assessed by feeling swelling of the common bile duct with the fingers. A clinical judgment was then made as to whether a bile leak was present on the transected surface of the liver. If a bile leak was found, the site was sutured on a point-by-point basis with an atraumatic needle.
In both groups, fibrin glue was spread over the transected surface. In the test group, the spread was done after completion of the test. A silicone drain was then inserted along the transected surface and connected to a closed drainage system.1 In selected patients, the biliary pressure was measured during the bile leakage test, using a hydraulic manometer connected to the balloon catheter.
Prophylactic antibiotics were given for 3 days in all patients. If the discharge was serous and had a bilirubin level of less than 86 µmol/L (<5 mg/dL), the drains were loosened about 1 week after the operation and were removed during the following week. If bile leakage occurred or the discharge was infectious, the drain was left in place until the discharge became serous and had a bilirubin level of less than 86 µmol/L (<5 mg/dL).
The primary end point was the incidence of postoperative bile leakage. Bile leakage was defined as bilious drainage, with a bilirubin level of 86 µmol/L or more (≥5 mg/dL), that continued for more than 1 week. When bile leakage persisted for more than 1 month, the patient was defined as having a biliary fistula. The secondary end points were the duration of drainage; the length of the postoperative hospital stay; and assessments of drainage fluid, including the bilirubin level on postoperative days 2, 6, and 10, the daily average volume, and bacteriologic culture results on postoperative day 4. When 2 or 3 drains were placed in one patient, the highest value was used for analysis. Other postoperative complications, if present, were also recorded. The outcomes were measured by separate physicians (H.T. and K.K.) who were blinded to the study arm.
During the last 2 years, the rate of postoperative bile leakage was 10% for the 138 hepatic resections in our department. Although the rates reported previously have been 6% to 11%,2- 6 the definitions were not always presented in the literature and some would indicate an event of long-persisted bile leakage.4- 6 According to our new criterion, which was more stringent than that of others,2- 6 it was inferred that the complication rate would be higher than that given in the historical data. We, therefore, assumed that the complication rate in the control group would be 15% and the rate could be reduced to 5% by the bile leakage test. One hundred forty patients in each arm would be required to detect a difference, with a 2-tailed type I error of 5% and a statistical power of 80%. An interim analysis was scheduled 2 years after the commencement of the trial.
All analyses were performed on an intention-to-treat basis. Continuous variables were expressed as the mean + SD. Postoperative variables showed skewed distributions and were, therefore, expressed as the median and range. The t test and the Mann-Whitney test were used to compare the outcomes between the 2 groups. Discrete variables were compared using the χ2 test and the Fisher exact test. Differences at P<.05 were considered statistically significant.
Between October 25, 1996, and November 16, 1998, a total of 197 patients underwent hepatic resection for liver tumors. Ninety-four patients were excluded according to exclusion criteria: 66 underwent previous laparotomy (39, hepatectomy; 16, cholecystectomy; 8, gastrectomy; and 3, other procedures); 22, concomitant resection of other organs (11, colectomy; 5, gastrectomy; 3, splenectomy; and 3, other procedures); 5, biliary reconstruction; and 1, emergency operation. One hundred three patients were eligible and thus randomized; 52 were enrolled into the bile leakage test group and 51 into the control group (Figure 1). The baseline characteristics of the patients in the 2 groups are similar, except for platelet counts (Table 1).
In 1 patient assigned to the control group, hepatectomy was later found to be contraindicated due to widespread tumor growth; this patient was excluded from further analysis. There still were no significant differences in any operative variables between the 2 groups after this patient was excluded (Table 2). Overall, 48 patients (47%) had liver cirrhosis, and only 24 (24%) had normal livers.
All patients in the test group underwent the bile leakage test. The volume of isotonic sodium chloride solution injected through the cystic duct was about 10 mL. A bile leak was observed in 21 patients (41%) during the test, and a median of 1 site (range, 1-6 sites) was closed using interrupted sutures. Biliary pressures were measured in 4 patients during the test and were 28, 28, 30, and 36 cm H2O. A technical complication related to the bile leakage test occurred in 1 patient: an injury of the common bile duct during insertion of the balloon catheter, which was successfully repaired by direct suturing. The patient's postoperative course was uneventful.
Bile leakage developed in 3 patients in the test group and in 2 in the control group (Table 3). The test did not influence the risk of bile leakage (odds ratio, 1.53; 95% confidence interval, 0.25-9.44). The drainage fluid investigations revealed no significant differences between the 2 groups. A transient bile leak that resolved within a week was observed in another 5 patients (2 in the test group and 3 in the control group). Between the 2 groups, there were no significant differences in the duration of drainage and the length of the postoperative hospital stay.
Operative morbidity rates were 29% in the test group and 24% in the control group (P = .50) (Table 4). There was no operative mortality. Relaparotomy was performed in 1 patient in the test group who was diagnosed as having developed a subphrenic abscess; however, there was no apparent abscess in the abdominal cavity. None of the other patients had a collection at the liver stump and underwent percutaneous drainage.
All 5 patients with bile leakage were treated conservatively (Table 5). Two of them showed delayed bile leakage, with an infection of the drain. Three patients (2 in the test group and 1 in the control group) developed a biliary fistula. In all patients, bile leakage subsided spontaneously.
There are 4 recommended methods of preventing biliary complications after liver surgery: intraoperative cholangiography,9 spreading fibrin glue on the transected liver surface,10 assessing bile duct patency by injecting air under ultrasonographic monitoring,11 and a bile leakage test.1,2,5 In our department, fibrin glue has been used routinely as a hemostatic agent during hepatic surgery, and bile leakage tests with isotonic sodium chloride solution have been performed in patients undergoing major hepatectomy as a simple way of detecting bile leakage. Tests of the same kind with dye solutions have been adopted in several institutions, but we found no difference in the detectability, whichever solution might be used.
The purpose of the bile leakage test is to detect insufficiently closed stumps of the bile ducts on the transected liver surface by elevating the biliary pressure. In this study, 41% of the patients in the test group showed a bile leak with this test, and these leaks were repaired successfully; an effective prevention of postoperative bile leakage seemed to be achieved. However, there were no significant differences in the incidence of the complication or of any of the other variables between the 2 groups, suggesting no efficacy of the test on prevention of bile leakage.
Including the transient cases, 5 patients showed bilious drainage despite the test. In 4 of them, a bile leak appeared on the day after the operation. What could be the cause of such bile leaks? One possibility is bile leakage from a separated bile duct with no communication to the main biliary tree. In nonanatomical hepatic resection, interruption of a bile duct can occur during parenchymatous division. There might be also peripheral biliary obstruction with minute stones or unexpected ischemic change causing interference with the bile duct patency. A bile leakage test cannot always examine all the biliary stumps on the transected liver surface. In this study, however, only 1 patient developed bile leakage after the nonanatomical resection (Table 5). Another possibility is bile oozing from crushed and exposed liver parenchymatous cells. A short period of examination cannot detect bile oozing at the capillary level. Such a microscopic bile leak might occur in some transient cases. On the other hand, a delayed bile leak is likely to be related to infection.12,13 Our study included 2 patients with a delayed bile leak (on postoperative days 6 and 10), and both of them also had a complication of drain infection (Table 5). Infectious irritation of the transected plane may cause collapse and dislodgment of fragile liver tissue and ensuing exposure of small bile ducts. Small bilomas may be transformed into abscesses. Bile leakage with infection is apt to be prolonged and to develop into a biliary fistula.
All of the patients who developed bile leakage in this series were treated conservatively. Bile leaks from small biliary stumps with some communication to the main biliary tree will usually close spontaneously, with the restoration of peristalsis and papillary function. Bile leaks from the separated parts often involve concomitant interruption of portal or arterial blood supply and will also subside with associated partial atrophy of the liver.14- 16 If effective drainage is achieved, it is, therefore, usually sufficient to observe the patient conservatively as long as careful management of the drains against infection is maintained. We believe that surgical intervention is only required when a bile leak originates from injury to a major duct.
The question of whether a drain is required at all following liver resection is another problem. Some researchers17,18 have noted the unnecessity of prophylactic drainage after liver resection and the risk of ascending infection along abdominal drains. Fong et al18 conducted a randomized trial showing no difference in complication rate between the group with closed-suction drains and the group without a drain, but they also reported the need for postoperative percutaneous drainage in 8% and 18% of each group against symptomatic fluid collections. We used isotonic sodium chloride solution–sealed closed drains in all patients in this study, and none of them required percutaneous drainage, although drain infection occurred in 8 patients (8%).
Normal biliary pressure is considered to be between 10 and 15 cm H2O.19,20 Excessive bile duct pressure has been reported to be a possible cause of cholangiovenous reflux and cholangitis.21,22 Yoshimoto et al22 found that cholangiovenous reflux develops at biliary pressures higher than 22 cm H2O and that regurgitation of bacteria can occur at pressures higher than 30 cm H2O. We measured biliary pressure in 4 patients during the bile leakage test and found values of 28 to 36 cm H2O. Although none of the patients developed cholangitis in this study, the test might apply excessive pressure to the biliary tree. We used palpation of the common bile duct to ascertain that the biliary tree was filled enough with isotonic sodium chloride solution. It may be preferable to standardize the pressure added on the bile duct because the detection of bile leaks might depend on the degree of pressure in the test.
The total incidence of bile leakage was 5 (5%) of 103 patients in this trial, and the interim analysis showed an extreme similarity in the rate of this complication between the 2 groups (P = .99); the rate in the test group (6%) was even higher than that in the control group (4%). Assuming that our rate of bile leakage of 5% would be reproducible and that the test still would reduce the rate to one third (1.7%), 460 patients in each group are required to achieve unequivocal statistical validity. A study of this size is clearly outside the resources of a single institution. Furthermore, we should reconsider the risk of the bile leakage test itself. To perform it, cholecystectomy is always necessary. Since an increased morbidity has been reported even in patients undergoing incidental cholecystectomy,23,24 additional cholecystectomy should not be justified if the efficacy to prevent the bile leakage is not suggested. Moreover, it was demonstrated that biliary pressure during the test seemed as high as the level at which cholangiovenous reflux might be caused. A technical complication related to the test occurred in 1 patient. We expected the usefulness of this test, with a 10% reduction in the rate of postoperative bile leakage, but the outcomes at the interim analysis have suggested that the test might not help in reducing the event. Although statistical validity has not been achieved yet, we decided to terminate the trial early based on the clinical point of view, as previously mentioned.
Despite the stringent criterion, the rate of bile leakage in this study was lower than the rate seen during the last 2 years in our department. Although only a few patients had no underlying liver disease (27%), the incidence of bile leakage in the control group of this study is the lowest in the reported literature.2- 6 The execution of a randomized trial may thus in itself have caused the surgeons to be more cautious in their attitude toward the management of bile leaks, and in other surgical manipulations, and this might have been responsible for reducing the incidence of bile leakage. Above all, meticulous management of the transected liver surface may be the important factor in minimizing postoperative bile leakage. This outcome was achieved in a high-volume center of liver surgery such as ours, but whether this may be reproducible by general surgeons needs to be evaluated.
In conclusion, the present study suggested no advantage in using a bile leakage test, and it should, therefore, not be routinely used during hepatic resection. A way of preventing postoperative bile leakage is meticulous management of the transected liver surface, rather than performing a bile leakage test.
Reprints: Masatoshi Makuuchi, MD, PhD, Division of Hepato-Biliary-Pancreatic and Transplantation Surgery, Second Department of Surgery, Faculty of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan (e-mail: firstname.lastname@example.org).