Daily volume of ascitic fluid drainage (A) and urinary output (B) postoperatively in patients with (solid circles and squares) and without (open circles and squares) a large volume of ascites. Circles and squares indicate median values; whiskers, limits of the 25th and 75th percentiles.
Serum albumin concentrations (A) and prothrombin times (B) before surgery and postoperatively in patients with (solid circles) and without (open circles) a large number of ascites. Circles indicate median values; whiskers, limits of the 25th and 75th percentiles. To convert serum albumin to grams per liter, multiply by 10.
Ishizawa T, Hasegawa K, Kokudo N, Sano K, Imamura H, Beck Y, Sugawara Y, Makuuchi M. Risk Factors and Management of Ascites After Liver Resection to Treat Hepatocellular Carcinoma. Arch Surg. 2009;144(1):46-51. doi:10.1001/archsurg.2008.511
Copyright 2009 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2009
To identify risk factors for a massive amount of ascites after liver resection to treat hepatocellular carcinoma and to evaluate our postoperative management strategy.
Two hundred three patients who underwent liver resection to treat hepatocellular carcinoma between January 1, 2003, and December 31, 2004.
Main Outcome Measures
Presence or absence of a large number of ascites (LA), defined as postoperative daily ascitic fluid drainage exceeding 10 mL per kilogram of body weight, and operative morbidity, mortality, and treatment costs.
A large number of ascites developed in 31 patients (15%). Multivariate analysis revealed that blood loss greater than 1000 mL (relative risk, 6.38; 95% confidence interval, 2.19-20.7; P = .001) and preoperative platelet count less than 100 × 103/μL (4.75; 1.75-13.1; P = .002) independently increased the risk of LA. In patients with LA, urinary output on postoperative days 1 to 3 was significantly lower than in patients without LA, and daily ascitic fluid volume tended to peak on postoperative day 7. No operative mortality was related to liver failure; however, patients with LA required a larger volume of fresh frozen plasma than those without LA (median [range], 1600 [0-16 800] mL vs 480 [0-5760] mL; P < .001), resulting in higher hospital costs.
Large blood loss and low platelet count were independent risk factors for LA. Although it was possible to safely manage postoperative ascites using routine administration of diuretic agents and fresh frozen plasma, step-by-step trials are required to reduce the need for transfusion of fresh frozen plasma.
Liver resection has been established as a safe and effective choice of treatment of hepatocellular carcinoma (HCC). As a result of advances in surgical techniques and perioperative management, operative mortality in high-volume medical centers has decreased to less than 5%.1- 7 However, postoperative morbidity in patients with HCC is still unsatisfactory at 26% to 56%,2,3,7- 9 probably because of decreased liver function. Ascites is one of the most common postoperative complications, with reported rates of 5% to 56%,1,3,5,7,8 and can lead to liver failure when large amounts of plasma are continuously lost in ascitic fluid. Thus, a strategy for managing ascites is needed to ensure the safety of resection of HCC.
Diuretic agents have been used widely to manage ascites after liver resection.5,10,11 Some authors have advocated postoperative administration of albumin12 or transfusion of fresh frozen plasma (FFP),5,10,12,13 probably to maintain circulating plasma volume and osmolarity in patients with a massive amount of ascites. However, to our knowledge, few reports have been published about the use of diuretic agents and blood products to manage ascites after liver resection. In our department, potassium canrenoate, one of the major metabolites of spironolactone, has routinely been administered perioperatively at dosages based on the indocyanine green retention rate at 15 minutes (ICGR15).3,14- 16 We have also advocated transfusion of FFP to replace the plasma volume and proteins lost in ascites.3,14- 16 The objectives of this study were to identify risk factors for massive ascites after liver resection to treat HCC and to evaluate our management strategy.
This study protocol was in accord with the Declaration of Helsinki and subsequent amendments. Subjects included 203 consecutive patients who had undergone liver resection to treat HCC at Tokyo University Hospital, Tokyo, Japan, between January 1, 2003, and December 31, 2004, after excluding 2 patients who had been receiving long-term hemodialysis.
Liver volume was estimated at computed tomography.17 The indications for surgery and operative procedure were determined based on the criteria consisting of 3 variables: (1) presence or absence of uncontrollable ascites, (2) serum bilirubin level, and (3) ICGR15.14 None of the patients had ascites before surgery.
Patients with at least 1 serum aspartate aminotransferase concentration greater than 100 IU/L (to convert to microkatals per liter, multiply by 0.0167) were treated with bed rest and intravenous injection of glycyrrhizin (Stronger Neo-Minophagen C; Minophagen Pharmaceutical Co, Ltd, Tokyo, Japan).3 When the ICGR15 value was more than 10%, potassium canrenoate was injected for 3 days before surgery to prevent postoperative sodium retention. The dose administered was based on the ICGR15 value as follows: 100 mg/d in patients with an ICGR15 of 10% to 19%, 200 mg/d in those with an ICGR15 of 20% to 29%, and 300 mg/d in those with an ICGR15 of 30% or greater.3,16
Anatomic resection18 of subsegment, Couinaud segment, sector, or hemiliver was the preferred operative procedure if the patient's liver functional reserve permitted. Liver transection was performed primarily using the clamp crushing method or temporarily with an ultrasonic dissector19 or an isotonic sodium chloride solution–linked radiofrequency coagulator20 during inflow occlusion. An abdominal drain was left with each cut surface and connected to a closed drainage system. Whenever thoracotomy was needed, a thoracic tube was also placed on a suction pump with a pressure of −20 cm H2O. During the operation, FFP was transfused at a rate that exceeded the amount of blood loss by 10% to 20%. Intraoperative red blood cell transfusions were given only if blood loss exceeded 1500 mL or the hematocrit reading decreased to less than 30%. Intravenous fluid infusion during surgery was set at a rate of 4.0 to 5.0 mL/kg/h.3,16
Red blood cells were transfused postoperatively only if the hematocrit reading after surgery was less than 20%.3,16 The volume of ascitic fluid drainage was measured daily, and the proteins and electrolytes lost in the fluid were replaced with FFP to maintain the total serum protein and serum albumin concentrations at 6.0 and 3.0 g/dL, respectively (to convert total serum protein to grams per liter, multiply by 10.0; albumin to grams per liter, multiply by 10).16 To avert sodium overload as a result of FFP transfusion, sodium-free crystalloids were used for maintenance intravenous infusion. Hypochloremic metabolic alkalosis induced by FFP transfusion was balanced by administering a chloride-rich amino acid solution at a dosage of about half the FFP transfusion.3,14 Infusion of a glucose solution was started at 0.1 g/kg/h and was increased by 0.05 g/kg/h every 24 hours.14 Patients with liver cirrhosis received a total fluid infusion of up to 40 mL/kg/d, and patients without liver cirrhosis received a total fluid infusion of up to 45 mL/kg/d.3,14,16
Potassium canrenoate was routinely administered at a dosage that depended on the ICGR15 value, as described earlier. Intravenous infusion of potassium canrenoate was changed to oral spironolactone at a ratio of 25 mg of spironolactone per 100 mg of potassium canrenoate. Furosemide was given if water tended to be retained or if potassium canrenoate alone was ineffective in controlling ascites. The dosage of furosemide was started at a ratio of 20 mg of furosemide per 100 mg of potassium canrenoate.16
Appropriate oral intake was resumed as soon as possible. The abdominal drains were left in place for at least 5 days and then were removed at a rate of 2 to 3 cm/d. The thoracic tube was removed when the amount of discharge decreased to less than 200 mL/d. Any symptomatic fluid collection in the chest or abdominal cavity was drained percutaneously under ultrasound guidance.3,16
We defined a large number of ascites (LA) as postoperative daily ascitic fluid drainage from thoracic and abdominal drains exceeding 10 mL/kg of preoperative body weight. We included body weight in the definition of LA because the significance of the plasma volume lost in ascitic fluid depends on the circulating plasma volume, which is correlated with body weight.21
Continuous data are expressed as median (range). Categorical data and continuous data are compared between the 2 groups using the Fisher exact test and the Wilcoxon rank sum test, respectively.
Background characteristics, surgery-related factors, and short-term results were compared between patients with LA (LA group) and those without LA (no-LA group). A multivariate analysis was performed to identify risk factors for LA. We included the following 8 potentially important preoperative and intraoperative factors in the analysis, considering their clinical significance: (1) positivity for hepatitis C virus antibody, (2) Child-Pugh class (A vs B), (3) ICGR15 (>20% vs <20%), (4) platelet count (≥100 × 103/μL vs <100 × 103/μL) (to convert platelets to ×109/L, multiply by 1.0), (5) creatinine clearance (≥50 mL/min/1.73 m2 vs <50 mL/min/1.73 m2) (to convert creatinine clearance to milliliters per second per meter square, multiply by 0.0167) before surgery, (6) extent of hepatectomy (≥1 sector vs >1 sector), (7) inflow occlusion time (≥60 minutes vs <60 minutes), (8) and blood loss (≥1000 mL vs <1000 mL). These 8 variables included some nonsignificant factors in the univariate analysis because any factors that are of potential importance can be incorporated into the model of the multivariate analysis whether they are statistically significant or not.22 Some significant factors were excluded from the analysis if they were highly correlated with 1 of the 8 variables. The results were expressed as adjusted odds ratios with 95% confidence intervals, and P values were calculated using the likelihood ratio test. Significance was defined as P < .05. Calculations were performed using commercially available software (JMP version 5.1.1; SAS Institute, Inc, Cary, North Carolina).
A large number of ascites developed in 31 of the 203 patients (15%) after liver resection to treat HCC. Table 1 lists the background characteristics of the patients with and without LA. Liver function was poorer in the LA group than in the no-LA group based on patient serum albumin concentration, platelet count, ICGR15, Child-Pugh class, and presence or absence of liver cirrhosis determined at pathologic examination. A larger percentage of patients in the LA group tested positive for hepatitis C virus antibody; they also had gastroesophageal varices diagnosed at preoperative upper gastrointestinal tract endoscopy.
The types of resection in the 2 groups were similar (Table 2). Blood loss and the percentage of patients who received red blood cell transfusions were higher in the LA group than in the no-LA group. The patients with LA required a larger volume of intraoperative FFP than did the patients without LA.
Multivariate analysis revealed that blood loss greater than 1000 mL (relative risk, 6.38; 95% confidence interval, 2.19-20.7; P = .001) and preoperative platelet count less than 100 × 103/μL (4.75; 1.75-13.1; P = .002) independently increased the risk of LA (Table 3). Figure 1A shows the trends in the volume of ascitic fluid drainage after resection. In the LA group, the daily volume of ascitic fluid drainage decreased on postoperative days 1 to 3 but began to increase on postoperative day 5 and peaked late after surgery (median, postoperative day 7; Table 4). Daily ascitic discharge in the no-LA group tended to decrease continuously. In the first 3 days after surgery, daily urinary output was significantly lower in the LA group than in the no-LA group (Table 4 and Figure 1B).
The total amount of potassium canrenoate given in the first 3 days after surgery was higher in the LA group than in the no-LA group (Table 4). Fresh frozen plasma was transfused in 182 of the 203 patients (90%) perioperatively. A total volume of postoperative FFP transfusion was larger in the LA group (median [range], 1600 [0-16 800] mL) than in the no-LA group (480 [0-5760] mL; P < .001).
Postoperative serum albumin concentrations were maintained at greater than 3.0 g/dL in most patients in both groups, although they were lower in the LA group than in the no-LA group on postoperative days 7 and 10 (3.1 [2.7-3.8] vs 3.3 [2.5-4.2] g/dL; P = .001, and 3.1 [2.8-3.9] vs 3.5 [2.6-4.1] g/dL; P = .002, respectively, Figure 2A). Postoperative prothrombin times in the 2 groups were similar (Figure 2B).
Postoperative morbidity was similar in the 2 groups except for the higher incidence of pleural effusion in the LA group (Table 4). Hyponatremia, defined as a serum sodium level less than 125 mEq/L (to convert to millimoles per liter, multiply by 1.0),23 did not develop in any patients in either group. An FFP-related complication, abrupt onset hypotension and tachycardia, occurred in 1 patient in the LA group during intraoperative FFP transfusion but was soon relieved by discontinuation of FFP and intravenous injection of hydrocortisol. The period required to remove the abdominal drains and the duration of the postoperative hospital stay were longer in the LA group than in the no-LA group. Total hospital costs and transfusion costs were higher in the LA group (Table 4). All patients recovered well except for 1 patient in the LA group who developed a pulmonary embolism from a tumor thrombus in the inferior vena cava during surgery and died of respiratory failure on postoperative day 50 as a result of aggressive progression of lung cancer.
A large number of ascites developed in 31 of the 203 patients (15%) who underwent liver resection to treat HCC. Multivariate analysis indicated that large blood loss and low preoperative platelet count independently increased the risk of LA. In the LA group, the daily ascitic fluid discharge showed a characteristic trend toward decreasing on postoperative days 1 to 3 and then increasing and peaking on postoperative day 7, and the daily urinary outputs in the first 3 days after surgery were significantly lower than those in the no-LA group.
Underlying mechanisms of a massive amount of ascites after liver resection is not well understood. Previous studies have revealed that in patients with cirrhosis, portal hypertension can trigger massive ascites by stimulating neurohormonal systems to promote renal water and sodium resorption.24,25 Similarly, high portal pressure after liver resection may cause insufficient urinary output in the early postoperative period and a subsequent increase in the volume of ascites. Among risk factors for LA identified in the present study, a platelet count of less than 100 × 103/μL indicates that portal hypertension already existed before surgey26 and that portal pressure can be elevated after liver resection because of a decrease in the hepatic vascular bed.27,28 A large amount of blood loss may also increase postoperative water and sodium retention because a decrease in effective arterial blood volume has been considered the major factor that promotes renal dysfunction in patients with portal hypertension.24,25 Another possible trigger for LA is the marked increase in splanchnic lymphatic flow as a result of oral feeding.29,30 Resumption of oral food intake, usually on postoperative day 3 in our series, may have accelerated the subsequent increase in ascitic fluid drainage in the LA group.
We managed the ascites after liver resection using routine intravenous injection of potassium canrenoate and transfusion of FFP to replace the plasma lost in the ascitic fluid.3,14- 16 Our management strategy is effective in preventing refractory ascites, which leads to liver failure. Meanwhile, the patients with LA required a larger amount of FFP transfusion and longer hospital stay, resulting in higher treatment costs. Thus, further improvement in the management of postoperative ascites is needed to prevent and treat massive ascites at lower cost.
Spironolactone and potassium canrenoate are the first-line drugs for diuretic therapy of ascites after resection of HCC because secondary hyperaldosteronism is the major factor that promotes renal sodium and water retention in patients with liver disease.23- 25 We routinely administered intravenous potassium canrenoate based on the ICGR15 values because the sodium and water retention tendency is correlated with the degree of liver cirrhosis. In our series, no patients experienced any adverse effects of potassium canrenoate therapy such as renal impairment, hyperkalemia, or hyponatremia. According to the recommendations23- 25 for the treatment of ascites in patients with cirrhosis, the dose of spironolactone and furosemide can be doubled to a maximum of 400 mg/d and 160 mg/d, respectively. However, it is unclear whether such higher doses of diuretic agents were equally safe in patients immediately after liver resection and were effective in preventing LA.
Management of massive ascites after liver resection often requires plasma volume expansion because continuous plasma loss in ascitic fluid can cause circulatory dysfunction and subsequent renal or hepatic impairment, as observed in patients with cirrhosis who undergo large-volume paracentesis without plasma volume replacement.23,24,31,32 Although the choice of volume expanders is a matter of controversy,23,24 some authors have shown that albumin is superior to synthetic plasma expanders in terms of a protective effect on the circulatory system after removal of LA.31,32 Fresh frozen plasma should be the best choice for plasma volume expansion in patients undergoing resection of HCC because it provides coagulation factors that should improve coagulation disorders caused by the underlying liver disease,33 which would temporarily be aggravated by the liver resection. Both the prothrombin times and serum albumin levels were well maintained after surgery in the present study, even in the LA group. The beneficial effect of FFP transfusion during and after liver resection is suggested by the zero mortality in our previous series,16 which was managed by using FFP in the same manner as in the present study, although to our knowledge, no data supporting the superiority of FFP over albumin in the management of ascites have been published.
The major disadvantage of FFP transfusion is the risk of transmitting diseases and adverse effects such as transfusion-related acute lung injury, acute allergic and anaphylactic reactions, and hemolysis.34,35 Acute allergic reactions are the most common complication (0.2-5.7 events per 10 000 transfusions),34,35 and 1 patient (0.5%) in our series had an acute allergic reaction. High cost is another problem posed by perioperative FFP use, and it can reduce the cost-effectiveness of resection to treat HCC. A prospective study is needed to establish stricter standards for FFP transfusion during and after liver resection or to substitute albumin or other plasma expanders for FFP without increasing the risk of postoperative liver failure after massive ascites.
In conclusion, large blood loss and low platelet count were independent risk factors for LA. Although it was possible to safely manage the ascites with routine administration of diuretic agents and FFP, step-by-step trials are essential to reduce the need for transfusion of FFP.
Correspondence: Norihiro Kokudo, MD, PhD, Hepato-Biliary-Pancreatic Surgery Division, Department of Surgery, Graduate School of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan (KOKUDO-2SU@h.u-tokyo.ac.jp).
Accepted for Publication: December 21, 2007.
Author Contributions: Dr Ishizawa had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Ishizawa, Hasegawa, and Makuuchi. Acquisition of data: Ishizawa, Sano, Imamura, and Sugawara. Analysis and interpretation of data: Ishizawa, Hasegawa, and Kokudo. Drafting of the manuscript: Ishizawa and Hasegawa. Critical revision of the manuscript for important intellectual content: Makuuchi. Obtained funding: Hasegawa. Study supervision: Kokudo, Beck, and Makuuchi.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grants from the Kanae Foundation for Life and Socio-Medical Science, the Public Trust Surgery Research Fund, the Japanese Clinical Oncology Fund and the Public Trust Haraguchi Memorial Cancer Research Fund in Japan and by grant 18790955, a Grant-in-aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Dr Hasegawa).