Risk factors for major morbidity after liver resection for hepatocellular carcinoma†
Presented to the 21st World Congress of the International Association of Surgeons, Gastroenterologists and Oncologists, Tokyo, Japan, November 2011
Bile leakage, and organ and/or space surgical-site infection (SSI) are common causes of major morbidity after partial hepatectomy for hepatocellular carcinoma (HCC). The purpose of this study was to analyse risk factors for major morbidity and to explore strategies for its reduction after partial hepatectomy for HCC.
Risk factors for bile leakage and organ/space SSI were analysed in patients who underwent partial hepatectomy for HCC between 2001 and 2010. The causes, management and outcomes of intractable bile leakage requiring endoscopic therapy or percutaneous transhepatic biliary drainage were analysed. In addition, causative bacteria, outcomes and characteristics of organ/space SSI were investigated. Risk factors were identified using multivariable analysis.
Some 359 patients were included in the analysis. The prevalence of bile leakage and organ/space SSI was 12·8 and 8·6 per cent respectively. Repeat hepatectomy and an operating time of at least 300 min were identified as independent risk factors for bile leakage. The main causes of intractable bile leakage were latent strictures of the biliary system caused by previous treatments for HCC and intraoperative injury of the hepatic duct during repeat hepatectomy. Independent risk factors for organ/space SSI were repeat hepatectomy and bile leakage. Methicillin-resistant Staphylococcus aureus was detected more frequently in organ/space SSI after repeat hepatectomy than after initial partial hepatectomy.
Repeat hepatectomy and prolonged surgery were identified as risk factors for bile leakage after liver resection for HCC. Bile leakage and repeat hepatectomy increased the risk of organ/space SSI. Copyright © 2012 British Journal of Surgery Society Ltd. Published by John Wiley & Sons, Ltd.
As a consequence of advances in surgical techniques and perioperative management, in-hospital mortality rates after partial hepatectomy for hepatocellular carcinoma (HCC) have improved greatly1–3. However, the still relatively high morbidity rates remain problematic, and bile leakage and organ/space surgical-site infection (SSI) continue to be common causes of major morbidity after partial hepatectomy for HCC4–10.
Various types of hepatectomy are now performed in many centres based on the degree of hepatic functional reserve and the location of the HCC. Anatomical liver resection, including subsegmentectomy, is reportedly effective for improving rates of overall and disease-free survival in patients with HCC compared with non-anatomical liver resection11. In addition, the rate of repeat hepatectomy for recurrent HCC has increased recently from 10 to 31 per cent as the prognosis for patients with HCC has improved12–14.
Anatomical and repeat liver resection for HCC have been performed when possible at the Department of Gastroenterological Surgery, Okayama University Hospital, Japan10, 14. The objective of this study was to investigate risk factors for bile leakage and organ/space SSI following partial hepatectomy for HCC in a large number of patients with a high proportion of anatomical or repeat liver resections. The causes, management and outcomes of intractable bile leakage and organ/space SSI were evaluated, and strategies to reduce major morbidity were also considered.
The medical records of patients who underwent partial hepatectomy without biliary reconstruction for HCC at the Department of Gastroenterological Surgery, Okayama University Hospital, between 1 January 2001 and 31 March 2010 were studied retrospectively. Repeat hepatectomy was considered if all tumours detected on preoperative imaging could be resected while leaving sufficient hepatic functional reserve. When recurrent HCCs were 2 cm or less in maximum diameter and three or fewer lesions were present, percutaneous ablation therapies were used despite the feasibility of repeat hepatectomy, depending on tumour location in the liver.
The type of laparotomy was decided on according to the type of partial hepatectomy and the patient's constitution. Preoperative cholangiography was not usually performed. Intraoperative ultrasonography was carried out to determine the extent of the HCC and the line of parenchymal transection. Parenchymal transection was done using an ultrasonic dissector (Sonop™ 5000; Aloka, Tokyo, Japan) combined with bipolar electrocautery. Glisson's pedicles in livers dissected by means of the ultrasonic dissector were ligated and small pedicles were resected using metal clips. For hemihepatectomies or extended resections, hilar dissection was performed to divide the ipsilateral branches of the hepatic artery and portal vein. The hepatic duct was exposed inside the liver during parenchymal transection and was ligated or oversewn using fine non-absorbable sutures. In hemihepatectomies or extended liver resections, parenchymal transection was usually performed without occlusion of vascular inflow. For segmentectomies or subsegmentectomies, Glisson's pedicles were transected at the hepatic hilus and an intermittent Pringle manoeuvre applied during parenchymal transection.
Intraoperative cholangiography was undertaken in selected patients if the integrity of the bile duct was in doubt. A bile leakage test using a cholangiography catheter was also performed in selected patients if a number of Glisson's pedicles were exposed in the plane of hepatic resection. In principle, two abdominal drainage tubes were positioned according to the type of hepatectomy. In hemihepatectomy, one drainage tube was placed on the cut surface of the liver and another at Winslow's foramen. In subsegmentectomy and segmentectomy, one drainage tube was placed on the cut surface of the liver and another in the right subphrenic space. From 2001 to 2005, an open drainage system was employed using 12-mm silicone Penrose drains (Kaneka, Osaka, Japan). From 2006 to 2010, a closed drainage system was used with 24-Fr BLAKE® silicone drains (Johnson & Johnson, Somerville, New Jersey, USA). Drains were removed when the drainage fluid was serous and contained no bile, which was usually around 5 days after surgery.
Definition of bile leakage and organ/space surgical-site infection
Postoperative bile leakage was defined as the drainage of macroscopic bile from surgical drains for more than 7 days after surgery. Major bile leakage was defined as macroscopic bile discharge exceeding 100 ml/day that did not decrease from one day to the next. Bile leakage was considered minor if it did not fulfil the definition for major bile leakage. Intractable bile leakage was defined as leakage requiring endoscopic retrograde biliary drainage (ERBD) or percutaneous transhepatic biliary drainage (PTBD) during postoperative management. SSIs were defined according to the National Infections Surveillance system15. Using these criteria, SSIs are classified as either incisional (superficial or deep) or organ/space. The definition of organ/space SSI was based on postoperative findings of at least one of the following: purulent drainage from a drain without macroscopic bile discharge; or intra-abdominal collection of purulent fluid confirmed at the time of reoperation or percutaneous computed tomography- or ultrasonography-guided drainage. Bile leakage was considered present if intra-abdominal fluid at the time of reoperation or percutaneous drainage contained macroscopic bile discharge. If purulent fluid was drained first and macroscopic bile leakage subsequently became apparent, this was defined as bile leakage. In contrast, if drainage of purulent fluid was still observed after the cessation of macroscopic bile leakage, organ/space SSI was considered present.
Prophylactic antibiotic regimens were as follows. At the initial partial hepatectomy, a first-generation cephalosporin was injected intravenously within 30 min before skin incision. In patients who underwent operations lasting longer than 3 h, additional antimicrobial agents were injected intravenously every 3 h, as recommended by the Centers for Disease Control guidelines15. These agents were also administered up to day 2 after operation. In repeat hepatectomy, a second-generation cephalosporin was injected intravenously as in the initial hepatectomy and continued until day 3.
Intervention for methicillin-resistant Staphylococcus aureus
With the exception of emergencies, all patients underwent preoperative evaluation for methicillin-resistant Staphylococcus aureus (MRSA), including nasal culture. In patients in whom MRSA colonization was detected from preoperative nasal cultures, decolonization was performed using intranasal mupirocin therapy, administered twice daily for 3–5 days before operation. Prophylactic intravenous infusion of vancomycin was not used in patients with intranasal MRSA colonization.
Analysis of risk factors for bile leakage and organ/space surgical-site infection
Patient demographics, preoperative liver function, operative and tumour factors were evaluated to determine their impact on the occurrence of bile leakage and organ/space SSI. Preoperative factors included patient age, sex, aetiology of liver disease, Child–Pugh grade, indocyanine green dye retention rate at 15 min (ICG-R15), serum albumin, history of diabetes mellitus, previous radiofrequency ablation (RFA) and previous transarterial chemoembolization (TACE). The cut-off level for ICG-R15 was set at 20 per cent, because an ICG-R15 of less than 20 per cent has been reported as safe for bisegmentectomy3, 4, 8. Tumour factors included the number of HCC lesions and the maximum diameter of individual HCCs. The cut-off level for HCC diameter was determined according to results from previous reports that analysed risk factors for morbidity after partial hepatectomy for HCC3, 4, 8, 10. Surgical factors evaluated were the type of partial hepatectomy, number of partial hepatectomies, duration of operation, duration of liver parenchymal transection, blood loss, need for blood transfusion and method of abdominal drainage. With regard to the type of partial hepatectomy, anterior segmentectomies and medial (S4) segmentectomies were subgrouped for analysis. Cut-off points for operating time and duration of liver parenchymal transection were determined by analysing the receiver operating characteristic (ROC) curve for bile leakage. The optimal cut-off for duration of operation was 302 min (sensitivity and specificity 0·690 and 0·313 respectively), and that for duration of liver parenchymal transection was 82 min (sensitivity and specificity 0·674 and 0·332). As a result, 300 min and 80 min were set as the cut-off levels for duration of operation and duration of liver parenchymal transection respectively.
Investigation of intractable bile leakage
Management and outcomes were investigated for patients with postoperative bile leakage. Indications for ERBD to treat postoperative bile leakage were based on postoperative findings of at least one of the following: amount of macroscopic bile discharge from surgical drains more than 200 ml/day at 2 weeks after surgery; amount of macroscopic bile discharge from surgical drains exceeding 100 ml/day at 4 weeks after surgery; or macroscopic bile discharge from surgical drains still continuing at 6 weeks after surgery. PTBD was indicated when postoperative cholangiography and biliary drainage by ERBD were considered impractical. The operative procedure, number of partial hepatectomies, timing of biliary drainage procedures, sites of bile leakage and possible causes of bile leakage were evaluated in patients with intractable bile leakage necessitating ERBD or PTBD.
Investigation of characteristics of prolonged hepatectomies
To clarify the clinical implications of a prolonged operation, patient demographics, preoperative liver function, tumour factors and surgical factors were compared between groups with an operation lasting less than 300 min versus at least 300 min.
Analysis of characteristics in organ/space surgical-site infection
Management and outcomes were investigated for patients with organ/space SSI. Organ/space SSI was classified according to the modified Clavien system16. The causative bacteria were identified for both incisional and organ/space SSIs. Preoperative and intraoperative parameters, causative bacteria and duration of postoperative hospital stay were compared between groups classified by the number of partial hepatectomies (initial versus repeat) in patients with organ/space SSI.
Continuous data are presented as mean(s.e.m.) unless indicated otherwise. Differences in qualitative variables were assessed using Fisher's exact test or the χ2 test, and quantitative variables were analysed using the Mann–Whitney U test. Univariable and multivariable logistic regression analyses were used to identify risk factors for bile leakage and organ/space SSI based on the 18 above-mentioned clinical factors. Variables with P≤0·050 in univariable analysis were included in the multivariable analysis. Relative risk was described by the estimated odds ratio (OR) with 95 per cent confidence interval. Two-sided P values were computed and P≤0·050 was considered statistically significant. All statistical analyses were performed using SPSS® II statistical software (IBM, Armonk, New York, USA).
A total of 359 patients were included in this analysis, comprising 292 men and 67 women (mean age 65 (range 32–89) years). The aetiology of liver disease in 332 patients with chronic hepatitis or liver cirrhosis was hepatitis C virus (163 patients), hepatitis B virus (122), both hepatitis C virus and hepatitis B virus (31) and alcoholic liver disease (16). Twenty-seven patients presented with HCC in a normal liver. Some 332 patients had Child–Pugh A and 27 had Child–Pugh B liver function. A total of 296 patients (82·5 per cent) underwent anatomical partial hepatectomy including subsegmentectomy. Repeat hepatectomy was performed in 59 patients (16·4 per cent). Laparotomy was performed through a -shaped incision in 287 patients, a Mercedes-type incision in 33, a midline incision in 23 and a thoracoabdominal incision in 16 patients. With the exception of two emergency cases, 357 patients underwent preoperative evaluation for MRSA using nasal culture. Nine (2·5 per cent) of the 359 patients showed colonization with MRSA on admission, and decolonization was carried out.
Risk factors for bile leakage
Bile leakage developed in 46 (12·8 per cent) of the 359 patients. Univariable analysis revealed several factors associated with an increased risk of developing bile leakage (Table 1). Repeat hepatectomy was associated with an increased risk of bile leakage, with an OR of 3·78 compared with the initial hepatectomy (P < 0·001). In contrast, neither previous RFA nor TACE had any significant impact on the occurrence of bile leakage. An operating time of at least 300 min was associated with an increased risk (OR 5·32; P < 0·001), as were blood loss of 2000 ml or more (OR 4·12; P < 0·001), blood transfusion (OR 3·66; P < 0·001) and duration of liver parenchymal transection (OR 3·97; P < 0·001). Multivariable analysis confirmed repeat hepatectomy and duration of operation at least 300 min as independent risk factors for bile leakage.
Table 1. Bacteria causing organ/space surgical-site infection
| Staphylococcus epidermidis||4||2|
| Enterococcus spp.||4||4|
| Streptococcus spp.||1||0|
| Escherichia coli||1||2|
| Klebsiella spp.||2||1|
| Pseudomonas spp.||2||4|
| Enterobacter spp.||1||1|
| Bacteroides spp.||0||1|
Risk factors for organ/space surgical-site infection
Organ/space SSI developed in 31 (8·6 per cent) of the 359 patients. Univariable analysis revealed several factors associated with an increased risk of developing organ/space SSI (Table 2). Repeat hepatectomy was associated with an increased risk of developing organ/space SSI, with an OR of 4·29 compared with initial hepatectomy (P < 0·001). In contrast, neither previous RFA nor TACE had any significant impact. The method of abdominal drainage (open Penrose drains or closed suction drains) had no significant influence. An operating time of at least 300 min was associated with an increased risk of organ/space SSI (OR 2·99; P < 0·001), as was duration of liver parenchymal transection (OR 5·15; P < 0·001), the presence of bile leakage (OR 3·16; P < 0·001) and blood loss of at least 2000 ml (OR 2·63; P < 0·001). Multivariable analysis confirmed repeat hepatectomy and presence of bile leakage as independent risk factors for organ/space SSI.
Table 2. Comparison between initial and repeat hepatectomies in patients with organ/space surgical-site infection
|Aetiology of liver disease|| || || |
| HCV + HBV||2||0||0·113|
|Child–Pugh grade|| || || |
| B||0||2|| |
|Diabetes mellitus|| || || |
| Yes||2||4|| |
|No. of HCC tumours|| || || |
| ≥ 1||3||9|| |
|Diameter of HCC (cm)*||4·5(0·9)||2·4(0·3)||0·039‡|
|Type of partial hepatectomy|| || || |
| Non-anatomical liver resection||0||3||0·104|
| Trisegmentectomy||0||0|| |
|Blood loss (ml)*||1833(511)||1697(307)||0·822‡|
|Duration of operation (min)*||333(11)||343(29)||0·767‡|
|Blood transfusion|| || || |
| Yes||5||9|| |
|Bile leakage|| || || |
| Yes||5||7|| |
|MRSA|| || || |
| Positive||2||10|| |
|Hospital stay (days)*||41(7)||63(11)||0·111‡|
Management and outcomes of bile leakage
In the 46 patients with bile leakage, minor leakage in 30 patients was managed successfully by conservative therapy comprising drainage alone in 23 patients and drainage with irrigation in seven.
Sixteen patients had major bile leakage. In eight of these patients, the leakage was treated using drainage with irrigation. One patient died from subsequent intractable ascites and liver failure during drainage with irrigation, whereas the other seven patients recovered. The remaining eight patients with major bile leakage needed either ERBD or PTBD. High-risk surgical procedures were performed in most of these patients and six underwent repeat hepatectomy. The median timing of biliary drainage procedures was 21·5 (range 2–45) days after surgery. Bile leakage sites identified on postoperative cholangiography included the hepatic duct in two patients and the raw surface of the liver in six. Possible causes of bile leakage as assessed by postoperative cholangiography were: stricture of the hepatic duct that was already present before surgery, possibly due to previous treatments for HCC in four patients (previous hepatectomies in 2 patients, previous TACE in 1 and previous RFA in 1), stricture of the hepatojejunostomy from previous pancreatoduodenectomy in one patient, dyskinesia of the papilla of Vater in one patient and intraoperative injury of the left hepatic duct during repeat hepatectomy in two patients. Three of these eight patients subsequently developed intractable ascites. In two patients, both bile leakage and intractable ascites were cured without intra-abdominal septic complications. The other patient, with a stricture and injury of the left hepatic duct caused by previous RFA, died from intractable ascites, uncontrollable biliary infection and liver failure. Bile leakage in the remaining five patients healed after either ERBD or PTBD, with no other major morbidities.
Characteristics of prolonged hepatectomies
Clinical parameters were compared between 126 hepatectomies that took at least 300 min and 233 procedures with a shorter operating time. HCC diameter was significantly greater in prolonged hepatectomies (5·6(0·4) versus 3·7(0·2) cm; P < 0·001). No significant differences were seen between groups in any other preoperative variables. Non-anatomical liver resection was performed significantly less frequent and repeat hepatectomy significantly more frequently in hepatectomies lasting at least 300 min. In addition, the duration of liver parenchymal transection was significantly longer in prolonged hepatectomies (63(1) versus 105(2) min; P < 0·001), blood loss was significantly greater (2176(194) versus 803(52) ml; P < 0·001) and the postoperative hospital stay was significant longer (38(3) versus 24(1) days; P < 0·001).
Management and outcome of organ/space surgical-site infection
Organ/space SSI in 31 patients was classified as follows: abscess on the cut surface of the liver in 26 patients, right subphrenic abscess in four patients and liver abscess in one patient. One patient was reoperated for a right subphrenic abscess, but died from a myocardial infarction. Eleven patients were treated successfully by percutaneous drainage of organ/space SSI. The infection in 19 patients healed with irrigation of the drain placed during surgery. Modified Clavien grades were: grade I, 0 patients; II, 13 patients; IIIa, 15 patients; IIIb, two patients; and IVa, one patient. There were no deaths associated with organ/space SSI in this series, but the postoperative hospital stay was significantly longer for patients with organ/space SSI than for those without (53(7) versus 27(1) days; P = 0·001).
Bacteria causing incisional and organ/space surgical-site infection
Causative bacteria for incisional and organ/space SSI comprised Gram-positive cocci in 17 and 19 patients, and Gram-negative rods in six and nine respectively, indicating similar proportions of Gram-positive cocci and Gram-negative rods in both incisional and organ/space SSI (Table 1). MRSA was the causative bacteria in 12 of 19 patients with organ/space SSI caused by Gram-positive cocci.
Comparison between initial and repeat hepatectomies in patients with organ/space surgical-site infection
HCC diameter was significantly larger in patients with organ/space SSI who underwent initial hepatectomy than in those who underwent repeat hepatectomy (Table 2). No significant differences were seen between groups in any other preoperative variables, including patient demographics and preoperative liver function. Nor were there any significant differences in operative parameters, including blood loss, duration of operation and blood transfusion. Rates of bile leakage were similar in the two groups. In terms of bacteria causing organ/space SSI, MRSA was more common in the repeat hepatectomy group than in the initial hepatectomy group.
Rates of bile leakage after partial hepatectomy for liver tumours and benign lesions range from 3·6 to 12·0 per cent5, 7, 9, 10, 17–21. However, no standardized definition of bile leakage after hepatectomy has been established. The International Study Group of Liver Surgery has proposed a consensus definition based on the postoperative course of bilirubin concentrations in serum and drainage fluid22. Application of a uniform definition of bile leakage is indispensable to enable standardized comparison of the results of different clinical reports and to facilitate objective evaluation of therapeutic modalities in the field of liver resection.
In the present study, a prolonged operating time and repeat hepatectomy were identified as independent risk factors for bile leakage, whereas the type of hepatectomy had no significant impact. Several groups reported that partial hepatectomies in which the cut surface exposed the major Glisson's sheath (central bisegmentectomy, S4 segmentectomy and S8 subsegmentectomy) were independent risk factors for bile leakage7, 20, 21. However, in the present study the standard types of partial hepatectomy were not risk factors for bile leakage. This was the case even if a wide cut surface with an exposed major Glisson's sheath was necessary, assuming liver function was appropriate and surgical procedures were performed carefully during transection of the liver parenchyma. A prolonged operation in this series was significantly associated with extended duration of liver parenchymal transection and/or repeat hepatectomy.
RFA for HCC located in the hepatic hilar region or TACE in which embolization of the feeding artery is necessary from the level of the left or right hepatic artery can induce ischaemic damage in the hepatic duct wall, possibly leading to a latent stricture of the hepatic duct. Anatomical liver resection, which requires exposure of the major Glisson's sheath, also has the potential to induce strictures of the hepatic duct early and/or late after hepatectomy. The present study revealed preoperative latent stricture of the biliary tree as the main cause of intractable bile leakage requiring invasive treatment. Preoperative assessment of biliary anatomy should thus be considered for patients who have undergone RFA for HCC located adjacent to Glisson's capsule in the hilar region, those who have had TACE in which embolization of the feeding artery was undertaken from the level of the left or right hepatic artery, and patients undergoing anatomical liver resection in which the cut surface exposes the major Glisson's sheath. In addition, various measures could be applied during surgery to diminish the incidence of major and intractable bile leakage. First, intraoperative cholangiography should be considered, particularly during repeat hepatectomies and in patients who have been treated with RFA or TACE for HCC located in the hepatic hilar region, as the identification of bile duct injury or stricture could allow immediate correction. Second, -tube drainage or transcystic duct drainage of the common bile duct should be considered in patients requiring decompression of the biliary tree, such as those with dyskinesia of the papilla of Vater. Third, particularly in repeat anatomical hepatectomies, division of the bile ducts should be performed inside the liver during parenchymal transection, as this may decrease the risk of injury to the bile ducts compared with division of the bile ducts at the liver hilum.
In the 1980s and 1990s, organ/space SSI after hepatectomy was reported as a fatal complication causing liver failure and death23, 24. Although organ/space SSI rates of 4·7–25 per cent after hepatectomy have been reported25–29, hospital mortality rates associated with organ/space SSI have declined6–8, 26, 28. Several groups have reported older age and presence of diabetes mellitus as independent risk factors for organ/space SSI26, 27. However, these variables were not identified as independent risk factors for organ/space SSI in the present study. Original results in this study were the identification of repeat hepatectomy as an independent risk factor for organ/space SSI and the frequent detection of MRSA in organ/space SSI after repeat hepatectomy for HCC.
It is assumed that most organ/space SSIs with MRSA after repeat hepatectomy develop as a result of contamination during repeat surgery by intra-abdominal colonization or microabscesses of MRSA formed after the initial partial hepatectomy. This assumption might be supported partially by present finding that the method of abdominal drainage (open or closed) had no significant influence on the occurrence of organ/space SSI. If this assumption is valid, preoperative interventions for MRSA, consisting of nasal culture and decolonization of nasal MRSA, will not greatly reduce the occurrence of organ/space SSI involving MRSA after repeat hepatectomy. Walsh and colleagues30 recently reported that an MRSA intervention programme, in which all patients received intranasal mupirocin and those colonized with MRSA received prophylactic intravenous infusion of vancomycin, resulted in near-complete and sustained elimination of MRSA SSIs after cardiac surgery. Regarding patients who undergo repeat hepatectomies, preoperative detection of intra-abdominal colonization or microabscess containing MRSA is difficult. MRSA intervention programmes thus need to be improved, particularly for patients undergoing repeat hepatectomies, probably by preoperative and/or postoperative prophylactic intravenous administration of anti-MRSA agents.
The present series of hepatectomies for HCC, which included a high proportion of anatomical and/or repeat hepatectomies, has shown the importance of establishing treatment strategies for preventing postoperative intractable bile leakage and for reducing rates of organ/space SSI due to MRSA after repeat hepatectomy.
The authors declare no conflict of interest.