Surgical resection of hepatocellular carcinoma. Post-operative outcome and long-term results in Europe: An overview


  • Daniel Jaeck,

    Corresponding author
    1. Centre de Chirurgie Viscérale et de Transplantation, Hôpital Universitaire de Hautepierre, Avenue Molière, Strasbourg Cedex, France
    • Centre de Chirurgie Viscérale et de Transplantation, Hôpital de Hautepierre, Avenue Molière, 67098 Strasbourg Cedex, France
    Search for more papers by this author
    • Telephone: 33-3-88-12-72-58; FAX: 33-3-88-12-72-86

  • Philippe Bachellier,

    1. Centre de Chirurgie Viscérale et de Transplantation, Hôpital Universitaire de Hautepierre, Avenue Molière, Strasbourg Cedex, France
    Search for more papers by this author
  • Elie Oussoultzoglou,

    1. Centre de Chirurgie Viscérale et de Transplantation, Hôpital Universitaire de Hautepierre, Avenue Molière, Strasbourg Cedex, France
    Search for more papers by this author
  • Jean-Christophe Weber,

    1. Centre de Chirurgie Viscérale et de Transplantation, Hôpital Universitaire de Hautepierre, Avenue Molière, Strasbourg Cedex, France
    Search for more papers by this author
  • Philippe Wolf

    1. Centre de Chirurgie Viscérale et de Transplantation, Hôpital Universitaire de Hautepierre, Avenue Molière, Strasbourg Cedex, France
    Search for more papers by this author


A multicenter retrospective review of 1467 patients treated by liver resection (LR) for hepatocellular carcinoma (HCC) in Europe over a 13-year period showed a mean mortality rate of 10.6%, which was correlated with the extent of LR, the etiology of cirrhosis and the study period with an improvement during the last years. Improved 5-year overall survival (20–51%) and disease-free survival (20–33%) reached similar rates in cirrhotic than in non-cirrhotic patients. Overall results were similar to those reported in Asian series as far as patients and tumor characteristics were comparable. (Liver Transpl 2004;10:S58–S63.)

Incidence of hepatocellular carcinoma (HCC) is rising in Western countries, including Europe.1, 2 HCC mainly occurs in patients with cirrhosis, which affects postoperative mortality and morbidity. Liver resection (LR) is usually considered as the standard option for resectable HCC while liver transplantation is the best option for small HCC in patients presenting with severe deterioration of liver function. Recently, more patients with HCC are being offered the option of surgical resection because of improved methods of early detection of HCC, of better patient selection, and improved perioperative management and surgical technique.3, 4 LR for HCC has become a safe procedure even in patients with liver cirrhosis and several European series were reported during the last decade with low mortality.5–21

The aim of this overview was to evaluate the postoperative outcome and long-term results after LR for HCC performed in Europe. The series collected here include a total of 1382 patients operated on in Europe.5–21 All the European series selected for this overview include at least 50 patients and were reported during the last decade.


HCC, hepatocellular carcinoma; LR, liver resection; AFP, α-fetoprotein; AST, asparate aminotransferase.

Patients and Methods

Resection of HCC in Europe

We collected the follow-up data from all the European series including at least 50 patients and published between 1990 and 2002. The latest series only from each institution were considered. We also reported data of our own series of 85 patients treated with LR for HCC in our institution from 1986 to 1999 (unpublished data). In these selected European series,11–21 analysis was focused on postoperative mortality (30-days), morbidity, tumor recurrences, and long-term survival.

Our Series

Between 1986 and 1999, 85 patients underwent LR for HCC in our institution. Forty-eight were cirrhotic and 37 had normal liver parenchyma. We included only cirrhotic patients with Child-Pugh score A and B. Patients with large tumors (> 5 cm) were selected for LR as in our institution liver transplantation is contraindicated for these tumors. For patients with alcoholic cirrhosis presenting small tumors (< 5 cm), LR was considered when they did not stop drinking alcohol. Patients over 70 years and those with associated diseases were selected for LR because of higher expected risk of postoperative mortality after liver transplantation. Patients presenting portal branch involvement and thrombosis were selected for LR to avoid the higher risk of tumoral recurrence after liver transplantation. Finally, patients with blood groups necessitating a long waiting time before transplantation were selected for LR to avoid progression of the tumor. The mean age was 63 ± 12 (range: 19–87). There were 68 male (80%) and 17 female (20%). Cirrhosis was alcoholic in 26 patients (54%), posthepatitic in 14 (29%), and from another or unknown etiology in 8 (17%). The majority (77%) of the patients were Child A. Fifty-three patients (62.3%) presented with tumor larger than 5 cm and 32 (37.7%) with tumor of 5 cm or less. Fifty patients (58.8%) presented with single HCC. The serum concentration of α-fetoprotein (AFP) was above 100 ng/ml in 39 patients (45.8%). Encapsulation was noticed in 66 patients (77.6%). Microscopic vascular invasion was detected in 21 patients (24.7%). Tumor was well-differentiated in 31 patients (36.4%), moderate in 37 (43.6%), and poor in 17 (20%). Hepatic transarterial chemoembolization was performed in 14 patients (16.4%). LR procedures included major hepatectomies in 31 patients (36.5%), bisegmentectomies in 20 (23.5%), monosegmentectomies in 28 (32.9%), and subsegmentectomies in 6 (7.1%). Major LRs in cirrhotic patients were exclusively performed in Child-Pugh A patients provided that the volume of the remnant liver was at least 50% of the whole non-tumoral liver volume. LR without inflow occlusion could be performed in 18 patients (21%). In 48 patients (56%) with normal liver parenchyma or Child-Pugh A cirrhosis, a continuous clamping was used when the required duration of hepatic inflow occlusion was less than 30 minutes. Finally, an intermittent clamping of the hepatic pedicle was used in 19 patients (23%) with an altered liver function and in those who needed pedicle clamping for more than 30 minutes (mean duration: 36 ± 19; range: 25–104). There were 3 postoperative deaths (3.5%) within 30 days after surgery. After a mean follow-up of 29 months, the 3- and 5-year overall survival rates were 44% and 38%, respectively. The 3- and 5-year disease-free survival rates were 41% and 33%, respectively.


Study Period and Patients Characteristics

In the collected series of LR for HCC, the mean duration of the study period was 10.8 ± 4.3 years (range: 5–19). A total of 1382 patients were operated on in 11 institutions and underwent 1413 LR for HCC (Table 1). Among them, 78% were male and their mean age was 59.4 ± 4.6 years.12, 13, 15–21 HCC was discovered by ultrasonography during routine follow-up of chronic liver disease in 669 asymptomatic patients (84.3%).11, 13, 15, 16, 20 The remaining 124 patients (15.7%) were symptomatic. Abdominal pain and discomfort, upper gastrointestinal bleeding, fever, abdominal mass, jaundice, and intraperitoneal bleeding were the most frequently presenting symptoms. In 4 European series, patients with normal liver parenchyma were included.12, 17, 18, 21 Among patients presenting chronic liver disease, the etiology of the underlying liver disease was alcoholic in 24% of the patients, viral in 57% (HCV 58% and HBV 42%) and from another cause in 19%. Eighty-three percent of the patients were considered as Child-Pugh A, 16 %B and 1% C. AFP serum level before operation was elevated above 100 ng/ml in approximately ⅓ of the patients.

Table 1. Selected Series in Europe Reporting Liver Resection for HCC
ReferenceStudy DesignStudy PeriodNumber of PatientsNumber of Liver Resection
  • *

    In these comparative studies the remaining patients underwent liver transplantation.

  • All studies were retrospective.

Franco et al. [11]Multicentric1983–19887272
Ringe et al. [12]Unicentric1974–1988192131*
Pitre et al. [13]Multicentric1984–1988153153
Bismuth et al. [14]Unicentric1980–199112060*
Di Carlo et al. [15]Unicentric1983–1992122122
Balsells et al. [16]Unicentric1987–19935153
Lise et al. [17]Unicentric1977–1995100101
Otto et al. [18]Unicentric1987–199610252*
Llovet et al. [19]Unicentric1989–199716477*
Grazi et al. [20]Unicentric1983–1998264264
Belghiti et al. [21]Unicentric1990–1999300328
Own seriesUnicentric1986–19998585

Tumor Characteristics

Single tumor was reported in 85% of the patients and multiple in 15%.11, 12, 14–17, 19–21 Tumor size measured on the specimen was 5 cm or less in 54% of the patients and more than 5 cm in 46%. Tumor encapsulation was observed in 68% of the patients (range: 52–79%).11, 13, 15, 16, 21 Forty-four percent of the patients presented with vascular invasion (range: 14–78%).12, 15, 16, 18, 19, 21 Histological grading and differentiation were evaluated in 766 patients: 79% of them presented with grade I or II.12, 15, 16, 19, 21 Details of patients and tumor characteristics are shown in Table 2.

Table 2. Patients and Tumor Characteristics in European Series
Franco [11] (n = 72)Ringe [12] (n = 131)Pitre [13] (n = 153)Bismuth [14] (n = 60)Di Carlo [15] (n = 122)Balsells [16] (n = 51)Lise [17] (n = 100)Otto [18] (n = 52)Llovet [19] (n = 77)Grazi [20] (n = 264)Belghiti [21] (n = 300)Own Series (n = 85)
Patients characteristics            
 Gender M/F89/42128/2597/2549/486/1436/1648/29216/48239/6168/17
 Age (mean, yr)6248616463605861625663
 Symptomatic HCC1851415641810
 Chronic liver disease7243153601225178257726424748
 Etiology of cirrhosis            
  Viral hepatitis313742345634516320016514
  Other or unknown303535482745718
 Child-Pugh score            
 Preoperative AFP level            
  < 100 ng/ml3874329537 (<10)72108 (≤20)16446
  > 100 ng/ml3432302758 (>10)5118 (>20)6439
 Preoperative TACE304614
Tumor characteristics            
 Tumor size            
  ≤ 5 cm50188246823749195816914232
  > 5 cm2211371144014613269515853
 Number of nodule            
 Differentiation grade            
  Grade 1 + 210685426711268
  Grade 3 + 411279105817
 Tumor encapsulation            
 Vascular invasion            

Preoperative and Operative Procedures

Preoperative transarterial chemoembolization was used in 63 patients (13%).13, 18, 20 The extent of LR was not reported for 112 patients.14, 18 In the remaining 1270 patients,11–13, 15–17, 19–21 1301 LR for HCC were performed. Among these patients the extent of LR was as follows: 1012 (77.5%) minor LR including subsegmentectomies, monosegmentectomies, or bisegmentectomies, 278 (21.7%) major LR, and 11 (0.8%) bilobar resections. Only few authors reported their strategy concerning the use of pedicular clamping during parenchyma transection. In 3 European series,13, 15, 20 LR was achieved without vascular clamping in 40% to 46% of the cases. Intermittent Pringle maneuver and hemihepatic vascular occlusion were increasingly used to perform parenchyma transection in the more recent period. Intraoperative transfusion was administered in 61.5% of patients (range: 41–91%).13, 15, 16, 20

Postoperative Mortality

There were 142 deaths within 30 days after surgery. The mean mortality rate was 10.6 ± 5.9% (range: 3.5–21%). Postoperative mortality was correlated with: the extent of LR,21 the presence of cirrhosis,21 the etiology of cirrhosis,21 and the study period.20 Mortality rate was higher in cases of alcoholic cirrhosis. Patients died mainly from liver failure, sepsis, or variceal bleeding. Less than 10% of the patients died from other causes. The mean morbidity rate after LR for HCC was 38.5 ± 15.0% (range: 16–61%).11, 12–17, 20 The most common postoperative complications were ascites (range: 22–35%) and transient liver failure (range: 4.9–19%).13, 17, 20

Tumor Recurrences and Survival

After a mean follow-up of 26 ± 18 months, a majority of the patients developed recurrences after LR for HCC. The main cause of death during the follow-up period was tumor progression (74% of the deaths). Lethal complications of the underlying chronic liver disease such as bleeding and liver failure contributed in 17% of the deaths.19

Prognostic factors that were analyzed are listed in Table 3.11–22 Child A cirrhosis or absence of cirrhosis, low preoperative AFP blood level, tumor size less than 3 cm, absence of satellite nodules, well or moderate differentiation grade, absence of vascular invasion, tumor encapsulation, and adequate resection margin were associated with improved overall survival. Independent prognostic factors for disease-free survival were Child stage; glutamic-oxaloacetic transaminase, gamma-glutamyltransferase, and AFP levels; size and number of tumor nodules; absence of satellite nodules; size of resection margins; absence of vascular invasion; and preoperative chemoembolization.

Table 3. Overall and Disease-Free Survival and Prognostic Factors For Survival (S) or Tumoral Recurrence (R) in European Series
Franco [11] (n = 72)Ringe [12] (n = 131)Pitre [13] (n = 153)Bismuth [14] (n = 60)DiCarlo [15] (n = 122)Balsells [16] (n = 51)Lise [17] (n = 100)Otto [18] (n = 52)Llovet [19] (n = 77)Grazi [20] (n = 264)Belghiti [21] (n = 300)Own series (n = 85)
  • *

    Prognostic factors were analyzed in a recent report of the same institution [22].

  • Abbreviation: AST, asparate aminotransferase.

Follow-up (mo)1519.5172932353329
Overall survival            
Disease-free survival            
Prognostic factors            
 Preoperative TACER
 Bilirubin levelS
 Gamma GTRR
 Portal hypertensionSR
 Absence of cirrhosisSS
 Unresolved liver failureS
 Portal vein thrombosisS
 Child A scoreSSSRS
 AFP serum levelRRS
 Fibrolamellar HCC varietyS
 Number of HCCSRSRRRR*
 Satellite nodulesSRR
 Tumor gradeSRS
 Tumor sizeSRSSRR
 pTNM stageS
 Vascular invasionSSRRR*S
 Curative resectionSS
 Width of resection marginSR
 Extrahepatic diseaseS
 Extent of resectionS


During the last decade, several European institutions reported their results of LR for HCC in cirrhotic patients. Acceptable mortality and morbidity rates have been obtained with an improved survival. Short- and long-term results in these patients might be similar to those reported by other institutions in the world and particularly in Asia.

Recently, a single institution with large experience of resection of HCC in cirrhotic patients reported a 1.3% operative mortality rate in a recent period due to improved diagnostic imaging and sophisticated surgical technique.20 These results are similar to those reported respectively by Makuuchi et al.23 and Fan et al.24 (mortality less than 1%). Makuuchi et al.23 recommended the use of the indocyanine green test to select the patients for LR. In all the European series, however, patients were selected for LR according to the Child-Pugh classification11–18, 20, 21 or to Okuda classification.19 Further studies are needed to evaluate which are the best criteria for selecting patients for surgery.

The largest series including 300 patients (247 with chronic liver disease) showed a relationship between postoperative mortality and the presence of cirrhosis with a higher mortality rate in alcoholic patients and in patients with viral hepatitis C compared to patients with viral hepatitis B.21 Another study showed the correlation between postoperative mortality rate and the extent of LR. Indeed, recent results are better than those observed previously (7.2% between 1983 and 1988 vs. 1.9% between 1989 and 1992).15 These results reflect the impact of a learning curve and have been recently confirmed.15, 20, 21

The improved survival is reflected by the 3- and 5-year survival rates of 39% and 26%, respectively, when compared with the natural history of the disease with no survivors 3 years after the initial diagnosis.25 LR can offer to a group of patients a hope of cure with a 10-year survival rate of 13%.21

Several factors affect survival and HCC recurrence after LR. They include preoperative liver function, AFP level, etiology of underlying liver disease, tumor characteristics, and extent of LR (Table 3). However, other factors such as gender, age, presence of cirrhosis, tumor size and tumor encapsulation did not influence overall survival.21 Another study defined independent factors that can predict increased risk of recurrence.19 These factors are high preoperative serum level of AFP and of gamma-glutamyltransferase, multinodular HCC, low differentiation grade, presence of satellite nodules and of vascular invasion.

A minority of the patients who underwent LR presented with Child C cirrhosis. There were only 14 patients (1%). These findings raise unanimously the doubts about the benefit of resection in patients with poor liver function.

Grazi et al.20 showed an improved survival, a reduced intraoperative transfusion, and a shorter hospital stay after LR of HCC in the more recent period of their experience, which is partly due to a better preoperative selection of the patients.20

None of these series was prospective and 2 were multi-institutional.11, 13 Actually, these series were heterogeneous. First, they included different methodologies: some series reported resection alone, others were comparative and reported resection versus percutaneous ethanol injection or resection versus transplantation. Second, several differences were noticed in inclusion criteria. Indeed, some authors performed LR alone for small HCC while others included small and large HCC. Some studies included patients with only single HCC20 or with fibrolamellar HCC variant.12 For these reasons, interpretation of the results should be made cautiously.

Currently liver transplantation is considered as the standard treatment for small HCC in patients with cirrhosis or advanced liver disease. But the choice between LR and liver transplantation is still controversial. Liver transplantation carries the great advantage of treating both the tumor and the cirrhosis. Therefore, liver transplantation eliminates or decreases significantly the risk of de novo development of HCC. In fact, optimal candidates for LR are also those for whom liver transplantation offers the best results in term of recurrence and survival.

Among the surgical therapies, it has been demonstrated by Llovet et al.19 that surgical resection for HCC was associated with better results than liver transplantation owing to an increased waiting time and a growing incidence of patient drop-out. In this series, survival was calculated according to the intention-to-treat principle. However, we showed recently, in an intention-to-treat design, that liver transplantation could offer better results than LR even in Child A patients with small HCC provided that the policy of organ procurement allows shorter waiting time.26


Incidence of HCC in Europe is rising but still lower than that reported in the East. Patient and tumor characteristics are not different between Eastern and European patients. Currently, LR can be achieved safely with low mortality and with similar disease-free and overall survival in Europe compared to that observed in the rest of the world. Etiology of the underlying liver disease is an important factor influencing the mortality rate after surgical resection (alcohol and viral hepatitis C). Long-term survival is mainly correlated with tumor characteristics. Moreover, there is a clear improvement in the management of cirrhotic patients with HCC over the years, particularly in terms of postoperative mortality rate (< 5%) and disease-free survival rate after LR (40% at 3-year and 25% at 5-year), which were comparable to that observed in patients with normal liver parenchyma. To improve the results of LR for HCC, more effort should be made to detect HCC earlier by the mean of a careful follow-up. Finally, liver transplantation should probably be considered as the treatment of choice that can prolong life expectancy of patients with HCC.