Late-onset bile leakage after hepatic resection Masaki Kaibori, MD,a Junzo Shimizu, MD,b Michihiro Hayashi, MD,c Takuya Nakai, MD,d Morihiko Ishizaki, MD,a Kosuke Matsui, MD,a Yong Kook Kim, MD,b Fumitoshi Hirokawa, MD,c Yasuyuki Nakata, MD,d Takehiro Noda, MD,e Keizo Dono, MD,e Akinori Nozawa, MD,f Masanori Kwon, MD,a Kazuhisa Uchiyama, MD,c and Shoji Kubo, MD,f Osaka, Japan
Background. Postoperative bile leakage can be a serious complication after hepatic resection. Few studies have analyzed patients according to the time of onset of bile leakage. We analyzed differences between patients with early- and late-onset bile leakage after hepatic resection and assessed clinical characteristics and outcomes in patients with late-onset leakage. Methods. Between 2008 and 2010, 1,009 patients underwent hepatic resection at 4 participating university hospitals and 2 community hospitals. Fifty-two patients (5.1%) with postoperative bile leakage were divided into an early-onset group (3 times the serum total bilirubin level), intra-abdominal accumulation of bile confirmed by drainage, or demonstration of bile leakage on postoperative cholangiography, for >3 days.20 Bile leakage was defined as early onset if it occurred during the first 13 postoperative days (PODs), and late onset if it occurred from POD 14, because the mean postoperative hospital stay in patients without postoperative complications was 12–13 days. The 52 patients were divided into an early-onset group (n = 34) and a late-onset group (n = 18). Before operation, each patient underwent conventional liver function tests and measurement of the indocyanine green retention rate at 15 minutes. Patients were screened for hepatitis virus infection by measurement of hepatitis B virus surface antigen and anti-hepatitis C virus antibody. Operative procedures. Operative procedures were classified according to the Brisbane terminology proposed by Strasberg et al.21 Anatomic resection was defined as resection of the tumor together with the related portal vein branches and corresponding hepatic artery territory. Anatomic resection was classified as hemihepatectomy, extended hemihepatectomy, sectionectomy (resection of 2 Couinaud subsegments22), or segmentectomy (resection of one Couinaud subsegment). All nonanatomic resections were classified as limited resection, which were performed in patients with central or peripheral tumors and moderate liver dysfunction.23 The Cavitron ultrasonic surgical aspirator dissector (Valleylab, Boulder, CO) was used to transect the hepatic parenchyma. Intraoperative cholangiography, bile leakage tests, and placement of drainage catheters were performed according to the policy of each hospital. Intra-abdominal drainage catheters were placed routinely after hepatectomy at 4 of the 6 hospitals. The drainage catheter was removed when the drainage fluid was minimal and not bile stained. Clinical characteristics of patients with bile leakage. Clinical characteristics were compared between the early- and late-onset groups. The characteristics were grouped as patient-related factors, intraoperative factors, and postoperative factors. Patient-related factors included sex, age, hepatitis virus infection status, body mass index, comorbidities, liver function test results, and diagnosis. Intraoperative factors included the operative procedure,
Surgery January 2015
operative approach, extent of hepatic resection, concomitant cholecystectomy, method of parenchymal dissection, use of absorbable sutures, use of the Pringle maneuver during parenchymal dissection, use of fibrin glue, performance of a bile leakage test, transcystic placement of a bile duct drain, placement of an intra-abdominal drainage catheter, and period of prophylactic antibiotics. Postoperative factors included the diagnostic method and treatments for bile leakage, postoperative complications, mortality, administration of therapeutic antibiotics, and pathogens grown from infected bile. Statistical analysis. Continuous variables are presented as median values (range). The significance of differences between the early- and lateonset groups was assessed using the Chi-square or Mann-Whitney U test as appropriate. Adjusted odds ratios (ORs) with 95% CIs for late-onset bile leakage were calculated by multivariate logistic regression analysis that included all factors found to be significantly different between the earlyand late-onset groups on univariate analysis. RESULTS Preoperative clinical characteristics. Table I shows the preoperative characteristics of patients in the early- and late-onset groups. There were no differences between the 2 groups in sex, age, hepatitis virus infection status, body mass index, diabetes mellitus, liver cirrhosis, gallbladder disease, prothrombin time, platelet count, or serum concentrations of total bilirubin, albumin, aspartate aminotransferase, and alanine aminotransferase. The lateonset group had a lesser indocyanine green retention rate at 15 minutes and a greater proportion of patients with metastatic liver tumors than the early-onset group. Intra- and postoperative characteristics. The operative procedure performed, operative approach, extent of hepatic resection, concomitant cholecystectomy, use of absorbable sutures, use of the Pringle maneuver during parenchymal dissection, performance of a bile leakage test, and intraoperative detection of major bile leakage did not differ significantly between the 2 groups (Table II). The proportion of patients who underwent parenchymal dissection using the Cavitron ultrasonic surgical aspirator dissector with the VIO soft coagulation system24 was greater in the late-onset group than in the earlyonset group (P = .027). The proportion of patients who underwent placement of an intra-abdominal drain was less in the late-onset group than in the early-onset group (P = .0007). The proportions of patients who underwent application of fibrin glue to the
Kaibori et al 39
Surgery Volume 157, Number 1
Table I. Preoperative clinical characteristics of patients in the early- and late-onset bile leakage groups Characteristic
Early group (n = 34)
Late group (n = 18)
P value
Sex (male:female) Age (years)* HBV:HCV:alcohol:nonBC Body mass index (%)* Diabetes mellitus Liver cirrhosis Gallbladder stone or polyp ICGR15 (%)* Platelet count (104/mL)* Total bilirubin (mg/dL)* Albumin (g/dL)* Prothrombin time (%)* AST (IU/L)* ALT (IU/L)* Diagnosis Hepatocellular carcinoma Metastatic liver tumor Benign disease
25:9 70 (22–79) 2:13:4:15 22 (19–30) 12 (35) 15 (44) 5 (15) 12 (5–26) 18 (6–41) 0.6 (0.3–1.3) 3.9 (2.9–4.8) 96 (70–125) 27 (12–92) 26 (11–111)
12:6 65 (36–83) 3:6:0:9 23 (19–26) 4 (22) 6 (33) 1 (6) 10 (3–22) 20 (9–38) 0.7 (0.2–1.3) 3.9 (3.2–4.7) 100 (82–121) 27 (18–61) 24 (11–110)
.603y .294z .295y .689z .298y .451y .326y .025z .223z .944z .761z .451z .242z .699z
24 (71) 4 (12) 6 (18)
12 (67) 6 (33) 0 (0)
.049y
*Values are median (range). yChi-square test. zMann-Whitney U test. Values in parentheses are percentages unless otherwise indicated. ALT, Alanine aminotransferase; AST, aspartate aminotransferase; HBV, hepatitis B virus; HCV, hepatitis C virus; ICGR15, indocyanine green retention rate at 15 minutes; NBC, non-hepatitis B or C virus.
cut surface of the liver and transcystic placement of a bile duct drain were less in the late-onset group than in the early-onset group. The proportion of patients who underwent placement of both intra-abdominal drainage catheters and a transcystic duct drain was less in the late-onset group than in the early-onset group (P = .0014). Multivariate logistic regression analysis found that parenchymal dissection using the VIO system (OR, 19.61; 95% CI, 1.35–250.00; P = .029) and absence of transcystic duct drain (OR, 15.63; 95% CI, 2.08–166.67; P = .020) were independent predictors of late-onset bile leakage (Table III). Diagnosis and treatment of bile leakage, and clinical courses of patients. The definitive diagnosis of bile leakage was on median POD 4 (range, 3–12) in the early-onset group and POD 21 (range, 15–45) in the late-onset group (Table IV). The longest period between hepatectomy and the diagnosis of bile leakage was 45 days. Twenty-two of the 34 patients in the early-onset group and all 18 patients in the late-onset group developed intra-abdominal infection. The proportion of patients who developed intra-abdominal infection was greater in the lateonset group than in the early-onset group (P = .0041). An intra-abdominal drainage catheter was placed during hepatectomy in 30 of the 34 patients in the early-onset group (Fig 1). In 21 of these 30 patients,
postoperative bile leakage was diagnosed by the bile concentration in the drainage fluid and was treated using the drainage catheter. In the other 9 patients, postoperative bile leakage was treated by abdominal paracentesis for drainage of an intra-abdominal fluid collection on POD 8 (range, 2–36), because suction on the drainage catheter was not sufficient to drain the bile leakage. In the 4 patients who did not undergo intra-abdominal drainage catheter placement, postoperative bile leakage was diagnosed and treated by abdominal paracentesis on POD 9 (range, 8–12). Five patients in the earlyonset group underwent endoscopic biliary drainage (endoscopic retrograde biliary drainage or endoscopic nasobiliary drainage) to treat the bile leakage. In these 5 patients, the bile leakage resolved on POD 19 (range, 11–144). An intra-abdominal drainage catheter was placed during hepatectomy in 8 of the 18 patients in the late-onset group (Fig 2). In these 8 patients, postoperative bile leakage was diagnosed and treated by abdominal paracentesis on POD 18 (range, 14–45), because the drainage catheter had been removed before the onset of bile leakage. In the 10 patients who did not undergo placement of an intra-abdominal drainage catheter, postoperative bile leakage was diagnosed and treated by abdominal paracentesis on POD 22 (range, 15–31). Endoscopic biliary drainage
40 Kaibori et al
Surgery January 2015
Table II. Intra- and postoperative characteristics of patients Characteristic
Early group Late group (n = 34) (n = 18)
P value
Operative procedure Anatomic resection 21 (62) 11 (61) .963y Nonanatomic 13 (38) 7 (39) resection Operative approach Open 32 (94) 17 (94) .962y Laparoscopic 2 (6) 1 (6) Extent of hepatic resection More than 9 (26) 8 (44) .189y hemihepatectomy Less than 25 (74) 10 (56) hemihepatectomy Synchronous 25 (74) 14 (78) .736y cholecystectomy Method of parenchymal dissection CUSA + VIO systemx 10 (29) 11 (61) .027y CUSA + othersk 24 (71) 7 (39) Absorption thread during parenchymal dissection Used 34 (100) 17 (94) .165y Not used 0 1 (6) Pringle’s maneuver during parenchymal dissection Yes 22 (65) 9 (50) .560y No 12 (35) 9 (50) Hemostatic device using fibrin glue or seat Used 30 (88) 10 (56) .008y Not used 4 (12) 8 (44) Bile leakage test during the operation Yes 26 (76) 11 (61) .245y No 8 (24) 7 (39) Detection of major bile leakage Present 16 (62) 5 (45) .367y Absent 10 (38) 6 (55) Detention of biliary drainage using cystic tube Yes 20 (59) 2 (11) .0009y No 14 (41) 16 (89) Detention of abdominal drainage Yes 30 (88) 8 (44) .0007y No 4 (12) 10 (56) Both detention of biliary and abdominal drainage Yes 17 (50) 1 (6) .0014y No 17 (50) 17 (94) Administrational period 2 (1–5) 2 (1–8) .175z of prophylactic antibiotics (d)* *Values are median (range). yChi-square test. zMann–Whitney U test. xVIO soft-coagulation system (VIO300D ERBE Elektromedizin, Tubingen, Germany). The tip has a paddle-type electrode attached to a small tube that drips saline solution during coagulation to reduce the contact resistance between the electrode and the tissues. kIncludes microwave tissue coagulation, bipolar water jet, TissueLink, and harmonic scalpel systems. Values in parentheses are percentages unless otherwise indicated. CUSA, Cavitron ultrasonic surgical aspirator; POD, postoperative day.
was performed in 12 patients, but was ineffective in 4. Two of these 4 patients underwent second hepatic resection to treat refractory bile leakage, and the other 2 died of sepsis. In the surviving patients who underwent endoscopic biliary drainage, the bile leakage resolved on POD 86 (range, 29–342). The proportions of patients who underwent paracentesis and endoscopic biliary drainage (endoscopic retrograde biliary drainage or endoscopic nasobiliary drainage) were greater in the lateonset group than in the early-onset group (P #.0001 each). In addition, 2 patients in the late-onset group underwent a second hepatic resection. The proportion of patients who underwent invasive treatment, including abdominal paracentesis, endoscopic biliary drainage, or second hepatic resection, was greater in the late-onset group than in the earlyonset group (P < .0001). The time to resolution of bile leakage was greater in the late-onset group (75 days; range, 29–342) than in the early-onset group (19 days; range, 5–144; P < .0001). Pathogens grown from infected bile. The proportions of patients with gram-positive bacteria (Staphylococcus epidermidis, S capitis subspeo, and S aureus), gram-negative bacteria (Acinetobacter baumanni and Pseudomonas aeruginosa), anaerobic bacteria (Bacteroides thetaiotaomicron), and fungi (Candida albicans) in the infected bile did not differ significantly between the early- and late-onset groups; however, the proportion of patients with Enterococcus faecalis in the infected bile was greater in the lateonset group (8/18) than in the early-onset group (5/34; P = .019). DISCUSSION Postoperative bile leakage remains an important problem associated with serious complications, such as sepsis and liver failure.7 Among the 52 patients with postoperative bile leakage in this study, 2 required a second liver resection for refractory bile leakage and 2 died of sepsis. In 8 of the 18 patients in the late-onset group, postoperative bile leakage was diagnosed after removal of the intra-abdominal drainage catheter placed at the time of hepatectomy. In these 8 patients, bile leakage may have developed in the late phase after hepatic resection (late development group). In the other 10 patients, bile leakage may have developed in the early phase (within a few days) or late phase after hepatic resection. In these patients, detection of bile leakage may have been delayed because of delayed onset of symptoms or abnormal laboratory test results, or because of lack of information about bile leakage
Kaibori et al 41
Surgery Volume 157, Number 1
Table III. Results of multivariate analysis to identify risk factors for late-onset bile leakage after hepatic resection Variable
Odds ratio
95% CI
P value
ICGR15 (%) Hepatocellular carcinoma CUSA + VIO system in method of parenchymal dissection Hemostatic device using fibrin glue or seat Absence of biliary drainage using cystic tube Absence of abdominal drainage Absence of both biliary and abdominal drainages
0.81 0.10 19.61 0.06 15.63 1.30 0.04
0.66–1.01 0.02–1.01 1.35–250 0.01–1.01 2.08–166.67 0.13–13.33 0.01–5.00
.066 .079 .029 .071 .02 .83 .2
CI, Confidence interval; CUSA, cavitron ultrasonic surgical aspirator; ICGR15, indocyanine green retention rate at 15 minutes.
Table IV. Diagnosis and treatment of bile leakage, and clinical courses of patients Variable
Early group (n = 34)
Late group (n = 18)
P value
Definitive diagnosis of bile leakage (POD)* Diagnostic method and treatments Abdominal drainage of intraoperative insertion Paracentesis of abdominal fluid collection ENBD or ERBD Yes No Reoperation Yes No Postoperative intra-abdominal abscess Present Absent Postoperative complications Sepsis Intractable ascites or plural effusion Surgical site infection Pneumonia Mortality Administration of therapeutic antibiotics Used Not used Cured days for bile leakage (POD)*
4 (3–12)
21 (15–45)
Background. Postoperative bile leakage can be a serious complication after hepatic resection. Few studies have analyzed patients according to the time of onset of bile leakage. We analyzed differences between patients with early- and late-onset bile leakage after hepatic resection and assessed clinical characteristics and outcomes in patients with late-onset leakage. Methods. Between 2008 and 2010, 1,009 patients underwent hepatic resection at 4 participating university hospitals and 2 community hospitals. Fifty-two patients (5.1%) with postoperative bile leakage were divided into an early-onset group (3 times the serum total bilirubin level), intra-abdominal accumulation of bile confirmed by drainage, or demonstration of bile leakage on postoperative cholangiography, for >3 days.20 Bile leakage was defined as early onset if it occurred during the first 13 postoperative days (PODs), and late onset if it occurred from POD 14, because the mean postoperative hospital stay in patients without postoperative complications was 12–13 days. The 52 patients were divided into an early-onset group (n = 34) and a late-onset group (n = 18). Before operation, each patient underwent conventional liver function tests and measurement of the indocyanine green retention rate at 15 minutes. Patients were screened for hepatitis virus infection by measurement of hepatitis B virus surface antigen and anti-hepatitis C virus antibody. Operative procedures. Operative procedures were classified according to the Brisbane terminology proposed by Strasberg et al.21 Anatomic resection was defined as resection of the tumor together with the related portal vein branches and corresponding hepatic artery territory. Anatomic resection was classified as hemihepatectomy, extended hemihepatectomy, sectionectomy (resection of 2 Couinaud subsegments22), or segmentectomy (resection of one Couinaud subsegment). All nonanatomic resections were classified as limited resection, which were performed in patients with central or peripheral tumors and moderate liver dysfunction.23 The Cavitron ultrasonic surgical aspirator dissector (Valleylab, Boulder, CO) was used to transect the hepatic parenchyma. Intraoperative cholangiography, bile leakage tests, and placement of drainage catheters were performed according to the policy of each hospital. Intra-abdominal drainage catheters were placed routinely after hepatectomy at 4 of the 6 hospitals. The drainage catheter was removed when the drainage fluid was minimal and not bile stained. Clinical characteristics of patients with bile leakage. Clinical characteristics were compared between the early- and late-onset groups. The characteristics were grouped as patient-related factors, intraoperative factors, and postoperative factors. Patient-related factors included sex, age, hepatitis virus infection status, body mass index, comorbidities, liver function test results, and diagnosis. Intraoperative factors included the operative procedure,
Surgery January 2015
operative approach, extent of hepatic resection, concomitant cholecystectomy, method of parenchymal dissection, use of absorbable sutures, use of the Pringle maneuver during parenchymal dissection, use of fibrin glue, performance of a bile leakage test, transcystic placement of a bile duct drain, placement of an intra-abdominal drainage catheter, and period of prophylactic antibiotics. Postoperative factors included the diagnostic method and treatments for bile leakage, postoperative complications, mortality, administration of therapeutic antibiotics, and pathogens grown from infected bile. Statistical analysis. Continuous variables are presented as median values (range). The significance of differences between the early- and lateonset groups was assessed using the Chi-square or Mann-Whitney U test as appropriate. Adjusted odds ratios (ORs) with 95% CIs for late-onset bile leakage were calculated by multivariate logistic regression analysis that included all factors found to be significantly different between the earlyand late-onset groups on univariate analysis. RESULTS Preoperative clinical characteristics. Table I shows the preoperative characteristics of patients in the early- and late-onset groups. There were no differences between the 2 groups in sex, age, hepatitis virus infection status, body mass index, diabetes mellitus, liver cirrhosis, gallbladder disease, prothrombin time, platelet count, or serum concentrations of total bilirubin, albumin, aspartate aminotransferase, and alanine aminotransferase. The lateonset group had a lesser indocyanine green retention rate at 15 minutes and a greater proportion of patients with metastatic liver tumors than the early-onset group. Intra- and postoperative characteristics. The operative procedure performed, operative approach, extent of hepatic resection, concomitant cholecystectomy, use of absorbable sutures, use of the Pringle maneuver during parenchymal dissection, performance of a bile leakage test, and intraoperative detection of major bile leakage did not differ significantly between the 2 groups (Table II). The proportion of patients who underwent parenchymal dissection using the Cavitron ultrasonic surgical aspirator dissector with the VIO soft coagulation system24 was greater in the late-onset group than in the earlyonset group (P = .027). The proportion of patients who underwent placement of an intra-abdominal drain was less in the late-onset group than in the early-onset group (P = .0007). The proportions of patients who underwent application of fibrin glue to the
Kaibori et al 39
Surgery Volume 157, Number 1
Table I. Preoperative clinical characteristics of patients in the early- and late-onset bile leakage groups Characteristic
Early group (n = 34)
Late group (n = 18)
P value
Sex (male:female) Age (years)* HBV:HCV:alcohol:nonBC Body mass index (%)* Diabetes mellitus Liver cirrhosis Gallbladder stone or polyp ICGR15 (%)* Platelet count (104/mL)* Total bilirubin (mg/dL)* Albumin (g/dL)* Prothrombin time (%)* AST (IU/L)* ALT (IU/L)* Diagnosis Hepatocellular carcinoma Metastatic liver tumor Benign disease
25:9 70 (22–79) 2:13:4:15 22 (19–30) 12 (35) 15 (44) 5 (15) 12 (5–26) 18 (6–41) 0.6 (0.3–1.3) 3.9 (2.9–4.8) 96 (70–125) 27 (12–92) 26 (11–111)
12:6 65 (36–83) 3:6:0:9 23 (19–26) 4 (22) 6 (33) 1 (6) 10 (3–22) 20 (9–38) 0.7 (0.2–1.3) 3.9 (3.2–4.7) 100 (82–121) 27 (18–61) 24 (11–110)
.603y .294z .295y .689z .298y .451y .326y .025z .223z .944z .761z .451z .242z .699z
24 (71) 4 (12) 6 (18)
12 (67) 6 (33) 0 (0)
.049y
*Values are median (range). yChi-square test. zMann-Whitney U test. Values in parentheses are percentages unless otherwise indicated. ALT, Alanine aminotransferase; AST, aspartate aminotransferase; HBV, hepatitis B virus; HCV, hepatitis C virus; ICGR15, indocyanine green retention rate at 15 minutes; NBC, non-hepatitis B or C virus.
cut surface of the liver and transcystic placement of a bile duct drain were less in the late-onset group than in the early-onset group. The proportion of patients who underwent placement of both intra-abdominal drainage catheters and a transcystic duct drain was less in the late-onset group than in the early-onset group (P = .0014). Multivariate logistic regression analysis found that parenchymal dissection using the VIO system (OR, 19.61; 95% CI, 1.35–250.00; P = .029) and absence of transcystic duct drain (OR, 15.63; 95% CI, 2.08–166.67; P = .020) were independent predictors of late-onset bile leakage (Table III). Diagnosis and treatment of bile leakage, and clinical courses of patients. The definitive diagnosis of bile leakage was on median POD 4 (range, 3–12) in the early-onset group and POD 21 (range, 15–45) in the late-onset group (Table IV). The longest period between hepatectomy and the diagnosis of bile leakage was 45 days. Twenty-two of the 34 patients in the early-onset group and all 18 patients in the late-onset group developed intra-abdominal infection. The proportion of patients who developed intra-abdominal infection was greater in the lateonset group than in the early-onset group (P = .0041). An intra-abdominal drainage catheter was placed during hepatectomy in 30 of the 34 patients in the early-onset group (Fig 1). In 21 of these 30 patients,
postoperative bile leakage was diagnosed by the bile concentration in the drainage fluid and was treated using the drainage catheter. In the other 9 patients, postoperative bile leakage was treated by abdominal paracentesis for drainage of an intra-abdominal fluid collection on POD 8 (range, 2–36), because suction on the drainage catheter was not sufficient to drain the bile leakage. In the 4 patients who did not undergo intra-abdominal drainage catheter placement, postoperative bile leakage was diagnosed and treated by abdominal paracentesis on POD 9 (range, 8–12). Five patients in the earlyonset group underwent endoscopic biliary drainage (endoscopic retrograde biliary drainage or endoscopic nasobiliary drainage) to treat the bile leakage. In these 5 patients, the bile leakage resolved on POD 19 (range, 11–144). An intra-abdominal drainage catheter was placed during hepatectomy in 8 of the 18 patients in the late-onset group (Fig 2). In these 8 patients, postoperative bile leakage was diagnosed and treated by abdominal paracentesis on POD 18 (range, 14–45), because the drainage catheter had been removed before the onset of bile leakage. In the 10 patients who did not undergo placement of an intra-abdominal drainage catheter, postoperative bile leakage was diagnosed and treated by abdominal paracentesis on POD 22 (range, 15–31). Endoscopic biliary drainage
40 Kaibori et al
Surgery January 2015
Table II. Intra- and postoperative characteristics of patients Characteristic
Early group Late group (n = 34) (n = 18)
P value
Operative procedure Anatomic resection 21 (62) 11 (61) .963y Nonanatomic 13 (38) 7 (39) resection Operative approach Open 32 (94) 17 (94) .962y Laparoscopic 2 (6) 1 (6) Extent of hepatic resection More than 9 (26) 8 (44) .189y hemihepatectomy Less than 25 (74) 10 (56) hemihepatectomy Synchronous 25 (74) 14 (78) .736y cholecystectomy Method of parenchymal dissection CUSA + VIO systemx 10 (29) 11 (61) .027y CUSA + othersk 24 (71) 7 (39) Absorption thread during parenchymal dissection Used 34 (100) 17 (94) .165y Not used 0 1 (6) Pringle’s maneuver during parenchymal dissection Yes 22 (65) 9 (50) .560y No 12 (35) 9 (50) Hemostatic device using fibrin glue or seat Used 30 (88) 10 (56) .008y Not used 4 (12) 8 (44) Bile leakage test during the operation Yes 26 (76) 11 (61) .245y No 8 (24) 7 (39) Detection of major bile leakage Present 16 (62) 5 (45) .367y Absent 10 (38) 6 (55) Detention of biliary drainage using cystic tube Yes 20 (59) 2 (11) .0009y No 14 (41) 16 (89) Detention of abdominal drainage Yes 30 (88) 8 (44) .0007y No 4 (12) 10 (56) Both detention of biliary and abdominal drainage Yes 17 (50) 1 (6) .0014y No 17 (50) 17 (94) Administrational period 2 (1–5) 2 (1–8) .175z of prophylactic antibiotics (d)* *Values are median (range). yChi-square test. zMann–Whitney U test. xVIO soft-coagulation system (VIO300D ERBE Elektromedizin, Tubingen, Germany). The tip has a paddle-type electrode attached to a small tube that drips saline solution during coagulation to reduce the contact resistance between the electrode and the tissues. kIncludes microwave tissue coagulation, bipolar water jet, TissueLink, and harmonic scalpel systems. Values in parentheses are percentages unless otherwise indicated. CUSA, Cavitron ultrasonic surgical aspirator; POD, postoperative day.
was performed in 12 patients, but was ineffective in 4. Two of these 4 patients underwent second hepatic resection to treat refractory bile leakage, and the other 2 died of sepsis. In the surviving patients who underwent endoscopic biliary drainage, the bile leakage resolved on POD 86 (range, 29–342). The proportions of patients who underwent paracentesis and endoscopic biliary drainage (endoscopic retrograde biliary drainage or endoscopic nasobiliary drainage) were greater in the lateonset group than in the early-onset group (P #.0001 each). In addition, 2 patients in the late-onset group underwent a second hepatic resection. The proportion of patients who underwent invasive treatment, including abdominal paracentesis, endoscopic biliary drainage, or second hepatic resection, was greater in the late-onset group than in the earlyonset group (P < .0001). The time to resolution of bile leakage was greater in the late-onset group (75 days; range, 29–342) than in the early-onset group (19 days; range, 5–144; P < .0001). Pathogens grown from infected bile. The proportions of patients with gram-positive bacteria (Staphylococcus epidermidis, S capitis subspeo, and S aureus), gram-negative bacteria (Acinetobacter baumanni and Pseudomonas aeruginosa), anaerobic bacteria (Bacteroides thetaiotaomicron), and fungi (Candida albicans) in the infected bile did not differ significantly between the early- and late-onset groups; however, the proportion of patients with Enterococcus faecalis in the infected bile was greater in the lateonset group (8/18) than in the early-onset group (5/34; P = .019). DISCUSSION Postoperative bile leakage remains an important problem associated with serious complications, such as sepsis and liver failure.7 Among the 52 patients with postoperative bile leakage in this study, 2 required a second liver resection for refractory bile leakage and 2 died of sepsis. In 8 of the 18 patients in the late-onset group, postoperative bile leakage was diagnosed after removal of the intra-abdominal drainage catheter placed at the time of hepatectomy. In these 8 patients, bile leakage may have developed in the late phase after hepatic resection (late development group). In the other 10 patients, bile leakage may have developed in the early phase (within a few days) or late phase after hepatic resection. In these patients, detection of bile leakage may have been delayed because of delayed onset of symptoms or abnormal laboratory test results, or because of lack of information about bile leakage
Kaibori et al 41
Surgery Volume 157, Number 1
Table III. Results of multivariate analysis to identify risk factors for late-onset bile leakage after hepatic resection Variable
Odds ratio
95% CI
P value
ICGR15 (%) Hepatocellular carcinoma CUSA + VIO system in method of parenchymal dissection Hemostatic device using fibrin glue or seat Absence of biliary drainage using cystic tube Absence of abdominal drainage Absence of both biliary and abdominal drainages
0.81 0.10 19.61 0.06 15.63 1.30 0.04
0.66–1.01 0.02–1.01 1.35–250 0.01–1.01 2.08–166.67 0.13–13.33 0.01–5.00
.066 .079 .029 .071 .02 .83 .2
CI, Confidence interval; CUSA, cavitron ultrasonic surgical aspirator; ICGR15, indocyanine green retention rate at 15 minutes.
Table IV. Diagnosis and treatment of bile leakage, and clinical courses of patients Variable
Early group (n = 34)
Late group (n = 18)
P value
Definitive diagnosis of bile leakage (POD)* Diagnostic method and treatments Abdominal drainage of intraoperative insertion Paracentesis of abdominal fluid collection ENBD or ERBD Yes No Reoperation Yes No Postoperative intra-abdominal abscess Present Absent Postoperative complications Sepsis Intractable ascites or plural effusion Surgical site infection Pneumonia Mortality Administration of therapeutic antibiotics Used Not used Cured days for bile leakage (POD)*
4 (3–12)
21 (15–45)
