Randomized Controlled Trial of Transarterial Lipiodol Chemoembolization for Unresectable Hepatocellular Carcinoma

Chung-Mau Lo, Henry Ngan, Wai-Kuen Tso, Chi-Leung Liu, Chi-Ming Lam, Ronnie Tung-Ping Poon, Sheung-Tat Fan, and John Wong
This randomized, controlled trial assessed the efcacy of transarterial Lipiodol (Lipiodol Ultrauide, Laboratoire Guerbet, Aulnay-Sous-Bois, France) chemoembolization in patients with unresectable hepatocellular carcinoma. From March 1996 to October 1997, 80 out of 279 Asian patients with newly diagnosed unresectable hepatocellular carcinoma fullled the entry criteria and randomly were assigned to treatment with chemoembolization using a variable dose of an emulsion of cisplatin in Lipiodol and gelatin-sponge particles injected through the hepatic artery (chemoembolization group, 40 patients) or symptomatic treatment (control group, 40 patients). One patient assigned to the control group secondarily was excluded because of unrecognized systemic metastasis. Chemoembolization was repeated every 2 to 3 months unless there was evidence of contraindications or progressive disease. Survival was the main end point. The chemoembolization group received a total of 192 courses of chemoembolization with a median of 4.5 (range, 1-15) courses per patient. Chemoembolization resulted in a marked tumor response, and the actuarial survival was signicantly better in the chemoembolization group (1 year, 57%; 2 years, 31%; 3 years, 26%) than in the control group (1 year, 32%; 2 years, 11%; 3 years, 3%; P .002). When adjustments for baseline variables that were prognostic on univariate analysis were made with a multivariate Cox model, the survival benet of chemoembolization remained significant (relative risk of death, 0.49; 95% CI, 0.29-0.81; P .006). Although death from liver failure was more frequent in patients who received chemoembolization, the liver functions of the survivors were not signicantly different. In conclusion, in Asian patients with unresectable hepatocellular carcinoma, transarterial Lipiodol chemoembolization signicantly improves survival and is an effective form of treatment. (HEPATOLOGY 2002;35:1164-1171.) epatocellular carcinoma is one of the most common cancers worldwide, and the largest concentration of cases is in Asia.1 The prognosis is poor because curative resection with partial hepatectomy or liver transplantation is applicable to only a small proportion of patients.2,3 For patients with unresectable disease, the goal of palliative treatment is to prolong survival and to control symptoms. Transarterial Lipiodol (iodized oil) (Lipiodol Ultrauide, Laboratoire Guerbet, Aulnay-

From the Center for the Study of Liver Disease, University of Hong Kong Medical Center, Queen Mary Hospital, Hong Kong, China. Received July 31, 2001; accepted February 21, 2002. Address reprint requests to: Chung-Mau Lo, M.S., Department of Surgery, University of Hong Kong Medical Center, Queen Mary Hospital, Hong Kong, China. E-mail: chungmlo@hkucc.hku.hk; fax: (852) 28175475. Copyright 2002 by the American Association for the Study of Liver Diseases. 0270-9139/02/3505-0019$35.00/0 doi:10.1053/jhep.2002.33156 1164

Sous-Bois, France) chemoembolization4 combines the effect of targeted chemotherapy5,6 with that of ischemic necrosis induced by arterial embolization.7-9 Its efcacy in reducing tumor growth and in prolonging survival has been suggested in nonrandomized studies.10-16 However, 3 randomized, controlled trials17-19 from Europe and Africa using different regimens of antitumor agents have failed to show any signicant benet in survival when comparing patients treated with chemoembolization to untreated controls. These studies included non-Asian patients, and in the 2 trials from France, 3 quarters of the patients had alcoholic liver disease. There are ethnic differences in the behavior of hepatocellular carcinoma,20 and the validity of the results of these trials may not be applicable to Asian patients whose liver disease most commonly is related to chronic hepatitis B virus infection.1,21 In addition, the efcacy and adverse effects of the treatment may be related to the regimen and technique of

HEPATOLOGY, Vol. 35, No. 5, 2002

LO ET AL.

1165

chemoembolization. Hence, we conducted a randomized controlled trial to evaluate the regimen of transarterial Lipiodol chemoembolization in patients with unresectable hepatocellular carcinoma at our center.

Patients and Methods

Selection of Patients. Consecutive new patients with diagnoses of unresectable hepatocellular carcinoma that were based on histology, cytology, or persistently elevated serum -fetoprotein levels ( 400 ng/mL) with typical imaging ndings were considered for entry into this single-center, open-label, randomized trial. Patients were not included if they refused to participate or had 1 or more of the following criteria: poor hepatic function (presence of hepatic encephalopathy, ascites not controlled by diuretics, history of variceal bleeding within last three months, a serum total bilirubin level over 50 mol/L, a serum albumin level below 28 g/L, or a prothrombin time of more than 4 seconds over the control); serum creatinine level of over 180 mol/L; history of previous treatment for the tumor or acute tumor rupture; presence of extrahepatic metastasis or vascular contraindications to chemoembolization (hepatic artery thrombosis, main portal vein thrombosis or arteriovenous shunting); or poor performance status (Eastern Cooperative Oncology Group performance status rating22 grade 4). Randomization. Randomization to either treatment with chemoembolization (chemoembolization group) or symptomatic treatment (control group) was performed without stratication by drawing consecutively numbered sealed envelopes. The protocol was approved by the ethics committee of the Faculty of Medicine of the University of Hong Kong, and informed consent was obtained from each patient. Treatment Procedure. The control group received only treatment for symptoms and complications. Patients in the chemoembolization group underwent transarterial Lipiodol chemoembolization after a standard protocol. They were fasted overnight with intravenous uid infusion for hydration. Intravenous amoxicillin-clavulanic acid (1.2 g), mannitol (20 g), and tropisetron (5 mg) were given before the procedure. The femoral artery was catheterized under local anaesthesia. Hepatic arteriography and superior mesenteric arterial portovenography were performed to dene the size and locations of tumor nodules and to exclude occlusion of the main portal vein. The right or left hepatic artery feeding the tumor was superselectively catheterized. Using the pumping method,4 an emulsion was prepared by mixing cisplatin (1 mg/mL) with Lipiodol in a volume ratio of 1 to 1. Various

amounts of the emulsion, up to a maximum of 60 mL (containing 30 mg of cisplatin) were injected slowly under uoroscopic monitoring according to the size of the tumor and the arterial blood ow. The aim was to deliver a sufcient amount of the emulsion to the tumorous areas without retrograde ow. If the tumor involved both lobes of the liver or if superselective catheterization was not possible, the emulsion was injected into the proper hepatic artery distal to the origin of the gastroduodenal artery. This was followed by embolization with small gelatin-sponge (Spongostan; Ferrosan, Johnson & Johnson Medical Ltd., Skipton, England) pellets of 1 mm diameter mixed with 40 mg of gentamicin. After the procedure, oral amoxicillin-clavulanic acid (375 mg 3 times per day) and sucralfate (500 mg 4 times per day) were administered for 3 days. Discharge from the hospital was decided according to the clinical state. Chemoembolization was repeated every 2 to 3 months and would be withheld or discontinued whenever vascular contraindications, poor hepatic function (dened in section on selection of patients), severe adverse effects, or progressive disease developed. Assessment of Outcome. The primary outcome measure was survival calculated from the date of randomization. Secondary outcome measures included the tumor response, patient tolerance, and liver function. The patients were followed up monthly at the outpatient clinic. Adverse effects that developed within 2 weeks after each session of chemoembolization were recorded. Follow-up assessments, including serum biochemistry, serum -fetoprotein level, indocyanine green retention test,23 and computed tomography scan were repeated every 3 months. Tumor response was assessed on the basis of computed tomographic measurement of tumor size (determined as the product of the 2 largest diameters) and serum -fetoprotein. Complete response was dened as complete disappearance of tumor on imaging or normalization of serum -fetoprotein. Response was considered to be major if tumor size or serum -fetoprotein decreased by more than 50% of the baseline measurement and minor if the reduction was 50% or less but more than 25%. Stabilization corresponded to variations 25% of the initial value and progression to an increase of more than 25%. Objective response was dened as the sum of complete and major responses. Follow-up was continued through October 30, 2000. Statistical Analysis. The 2-year survival rate of the control group was estimated at 10% according to the published data on the natural course of Asian patients with untreated Okuda stage I and II hepatocellular carcinoma.24 Based on our previous case series,13 reporting a 38% 2-year survival in patients treated with the present

1166

LO ET AL.

HEPATOLOGY, May 2002

regimen of chemoembolization, 40 patients were needed in each group to have a level of statistical signicance of 5% and a power of 80%. Comparison between groups was made on an intention-to-treat basis. Continuous variables were expressed as median (interquartile range) and compared using the Mann-Whitney U test. Categorical variables were compared with the 2 test or Fishers exact test. Spearmans correlation coefcient and linear regression were used to express the relationship between the dosage of cisplatinLipiodol emulsion and tumor size. Univariate analysis for 13 baseline variables to identify predictors of survival was performed by estimating survival rate according to the Kaplan-Meier method25 and comparing with the use of the log-rank test. For continuous variables, the cutoff was set at the median value. In addition, a univariate Cox proportional hazards model was t individually for each continuous variable. The survival curves of the treatment and control groups were then compared with stratication according to each statistically signicant prognostic factor. Finally, all the signicant prognostic factors identied from univariate analysis were put into a Cox proportional hazards model26 for multivariate analysis. Statistical analysis was performed with the SPSS (SPSS Inc., Chicago, IL) computer software program.

Table 1. Baseline Characteristics of the Study Patients According to the Treatment Group
Chemoembolization (N 40) Control (N 39)

Age (yr)* Sex (men/women) Serum hepatitis B surface antigen (positive/negative) Serum creatinine ( mol/L)* Serum total bilirubin ( mol/L)* Serum albumin (g/L)* Serum alanine aminotransferase (1 /L)* Prothrombin time (s)* Serum -fetoprotein (ng/mL)* Serum -fetoprotein (ng/mL; 20/ 20-500/ 500) Indocyanine green retention at 15 minutes (%)* ECOG performance status rating (0/1/2/3) Presenting symptom (asymptomatic/ symptomatic) Diameter of largest tumor mass (cm)* Number of tumors (solitary/multinodular) Portal vein obstruction (right/left/ main) Okuda stage (I/II)

62 (53-69) 36/4 34/6 92 (82-102) 14 (10-21) 38 (32-42) 51 (38-83) 11.5 (10.9-12.4) 505 (55-5,874) 6/14/20 24 (6-33) 20/16/3/1 12/28 7 (4-14) 17/23 6/3/0 19/21

63 (53-70) 34/5 29/10 87 (78-98) 13 (11-23) 37 (33-40) 53 (35-88) 11.5 (10.8-12.5) 500 (58-24,458) 8/12/19 18 (6-38) 14/19/4/2 10/29 7 (5-11) 15/24 7/5/0 18/21

Results
From March 1996 to October 1997, we evaluated 387 Asian patients with newly diagnosed hepatocellular carcinoma. A total of 108 patients underwent surgery (partial hepatectomy, 104; cryotherapy, 4) and 199 of 279 patients (71.3%) with disease not amenable to surgery were not included in the study because of vascular contraindictions (53 patients), extrahepatic metastasis (26 patients), poor hepatic function (39 patients), refusal by patients (32 patients), previous treatment for the tumor (24 patients) or acute tumor rupture (8 patients), poor performance status (11 patients), indication for percutaneous intralesional ethanol injection (2 patients), and others (4 patients). The remaining 80 patients were allocated randomly to the chemoembolization group (40 patients) or the control group (40 patients). One patient assigned to the control group was excluded secondarily because of unrecognized pulmonary and bone metastases on computed tomography scan taken before randomization. He received conservative treatment and survived for 3 months. Thus, the data consisted of observations in 79 patients. Sixty-three patients (80%) had positive serology test results for hepatitis B surface antigen. Thirty-seven patients (47%) had Okuda stage I disease, and 42 (53%) had stage II disease. The 2 groups were well balanced in regard to the baseline characteristics (Table 1).

NOTE. P .05 for all variables when the 2 groups are compared. *Values are medians, with interquartile ranges shown in parentheses. Assessed by computed tomography.

Chemoembolization Treatment. Forty patients assigned to the chemoembolization group received a total of 192 courses of chemoembolization, with each patient receiving a median of 4.5 courses (range, 1-15). Two patients were scheduled for continuation of chemoembolization as of the date of the latest follow-up. The remaining 38 patients had treatment stopped because of progressive disease (12 patients), death (7 patients), poor liver function (6 patients), adverse effects (6 patients), patient refusal (3 patients), arteriovenous shunting (2 patients), and hepatic artery thrombosis (2 patients). Ninety-four courses (49%) of the chemoembolization were performed by selective injection into the right or left hepatic artery. The median volume of cisplatin-Lipiodol emulsion injected in 1 course was 20 mL (range, 2-60), and the dosage was signicantly related to the tumor size (r 0.70; P .001) (Fig. 1). The most common clinical adverse effect was a self-limiting syndrome consisting of fever, abdominal pain, and vomiting (Table 2). The median hospital stay for each course of treatment was 2 days (range, 1-21). Overall Survival. Two patients, 1 in each group, were lost and could not be contacted after a follow-up of

HEPATOLOGY, Vol. 35, No. 5, 2002

LO ET AL.

1167

Fig. 1. Correlation between volume of cisplatin-Lipiodol emulsion injected during the rst course of treatment and the tumor size (linear regression and Spearman correlation coefcient).

4 months and 9 months. These were treated as censored observations. At the time of the nal analysis, 31 patients of the chemoembolization group and 37 patients of the control group had died. The causes of death were tumor progression (20 in chemoembolization group and 34 in control group); hepatic failure (5 in chemoembolization group and 1 in control group); gastrointestinal bleeding (4 in chemoembolization group); sepsis (2 in control group); and rupture of tumor (2 in chemoembolization group). The actuarial survival was signicantly better in the chemoembolization group than in the control group (Fig. 2). The estimated 1-year, 2-year, and 3-year survival rates were 57%, 31%, and 26% for the chemoembolization group and 32%, 11%, and 3% for the control group (relative risk of death in the chemoembolization group was 0.50; 95% CI, 0.31-0.81; P .005). Univariate analysis with log-rank test identied 5 signicant prognostic factors: study treatment, presenting symptom, unilobar portal vein obstruction, tumor size, and Okuda stage (Table 3). Tumor size also was a signicant prognostic factor
Table 2. Complications of 192 Courses of Chemoembolization in 40 Patients Assigned to Receive Treatment
Complication No. of Complications (%)

when each continuous variable was put individually into a univariate Cox model (P .03). Comparison of survival between the chemoembolization and control groups stratied according to each of the other 4 prognostic factors revealed that the survival benet of chemoembolization was signicant in each subgroup except for those with tumors larger than 5 cm in diameter or unilobar portal vein obstruction (Table 4). With multivariate analysis, 2 of the 5 variables in the nal Cox model were statistically signicant: treatment with chemoembolization and unilobar portal vein obstruction. Patients who received chemoembolization had a relative risk of death of 0.49 (95% CI, 0.29-0.81; P .006) as compared with those of the control group, thus conrming the independent survival benet of treatment with chemoembolization. The relative risk of death for patients with unilobar portal vein obstruction was 2.71 (95% CI, 1.38-5.32; P .004). Secondary End Points. Tumor response on computed tomography could be determined in 46 patients (28 in the chemoembolization group and 18 in the control group) who survived more than 3 months and had a measurable tumor on computed tomographic scan. In the chemoembolization group, there was no complete response, 11 major responses, 6 minor responses, 7 stabilizations, and 4 progressions. In the control group, there was no complete response, 1 major response, 2 minor responses, 6 stabilizations, and 9 progressions. The rate of objective tumor response in measurable patients was signicantly higher in the chemoembolization group than in the control group (39% vs. 6 %; P .014). -Fetoprotein response could be assessed in 50 patients (29 in the chemoembolization group and 21 in the control group) who had a baseline serum -fetoprotein level 20 ng/mL and survived more than 3 months. In the chemoembolization group, there were 9 complete responses, 12 major responses, no minor response, 1 stabilization and 7 progressions. In the control group, there was no complete response, 2 major responses, no minor

Fever 38C Abdominal pain Vomiting Ascites Gastrointestinal bleeding Bleeding from femoral puncture site Encephalopathy Ruptured tumor Pleural effusion Liver abscess Hematuria Hypotension Bradycardia

63 (32.8) 50 (26.0) 32 (16.7) 10 (5.2) 8 (4.2) 3 (1.6) 3 (1.6) 2 (1.0) 2 (1.0) 1 (0.5) 1 (0.5) 1 (0.5) 1 (0.5)

Fig. 2. Probability of survival in patients treated with chemoembolization and in patients of the control group (log-rank test, P .002).

1168

LO ET AL.

HEPATOLOGY, May 2002

Table 3. Univariate Analysis of Prognostic Variables for Survival

Probability of Survival (%) No. of 12 Patients Months 24 36 Months Months

Characteristics

chemoembolization, the median number of admissions for adverse events was 1 (range, 0-8) in both the chemoembolization group and the control group for a cumulative median of 13 days (range, 0-215) and 13 days (range, 0-96), respectively (P .31).

Study treatment Chemoembolization Control Sex Men Women Age (yr) 60 60 ECOG performance status rating 0 1-3 Presenting symptom Asymptomatic Symptomatic Tumor size (cm) 5 5 Tumor number Single Multiple Unilobar portal vein obstruction Negative Positive Okuda stage I II Serum albumin (g/L) 37 37 Serum bilirubin ( mol/L) 14 14 Indocyanine green retention at 15 minutes (%) 20 20 -Fetoprotein (ng/mL) 500 500

40 39 70 9 34 45

57 32 44 56 35 52

31 11 20 33 15 26

26 3 15 11 12 17

.002

Discussion
Hepatocellular carcinoma is a common form of malignancy in Asia. In some Western countries, a high incidence rate frequently is seen among Asian immigrants as well.1 The causal factors in Asian patients are distinctly different from those in western patients, and chronic hepatitis B infection is the most common etiology. The treatment for hepatocellular carcinoma remains difcult because of the advanced tumor stage at diagnosis and the associated cirrhosis. No single available therapy is applicable to all patients and the treatment should be catered according to the clinical state of the individual patient. Hepatic resection offers a chance of cure for a minor proportion of patients with early tumor and preserved liver functions. Because of the shortage of organ donors, the role of liver transplantation remains limited. The majority of the patients with unresectable hepatocellular carcinoma are treated by various palliative therapies. In considering treatment for these patients with variable tumor stages and liver reserve, it is important to select those who may potentially benet from the procedure. In the present study, advanced tumor with extrahepatic metastases or portal vein thrombosis (the most common vascular contraindication) and poor hepatic function were the 2 main reasons for exclusion because these patients would denitely not benet from the study treatment. Hence, less than 30% of the patients with disease not amenable to surgery could be included, and our results showed that

NS (.720)

NS (.274)

34 45 22 57 26 50 32 47

43 46 77 33 65 37 55 39

18 23 43 13 39 13 29 16

9 19 29 9 27 9 16 14

NS (.384)

.004

.019

NS (.225)

58 21 37 42 40 39 41 38

60 5 61 31 44 46 41 49

27 5 35 10 16 27 23 19

18 5 23 7 10 19 18 11

.001

.003

NS (.341)

NS (.967)

38 41 40 39

39 50 54 36

20 23 32 10

17 13 19 10

NS (.635)

NS (.059)

Table 4. Comparison of Survival Between the Chemoembolization and Control Groups Stratied by Baseline Prognostic Variables
Chemoembolization Control P

response, 1 stabilization and 18 progressions. The rate of objective response was signicantly higher in the chemoembolization group than in the control group (72% vs. 10%; P .001). Extrahepatic metastasis was found in 11 patients in the chemoembolization group and 5 patients in the control group (P .105). Apart from a lower serum bilirubin level in the chemoembolization group at 3 months (P .038), there was no signicant difference in the liver function of the survivors of the 2 groups, as assessed by serum bilirubin level, serum albumin level, and indocyanine green retention rate (Table 5). Excluding the scheduled hospital admissions for

Presenting symptom Asymptomatic Symptomatic Unilobar portal vein obstruction Negative Positive Tumor size (cm) 5 5 Okuda stage I II

25.4 (17.5) 11.2 (2.6) 18.0 (3.5) 5.1 (2.2) 29.8 (12.2) 11.2 (1.8) 25.4 (9.1) 9.2 (4.1)

16.6 (2.5) 5.2 (1.4) 9.2 (5.6) 2.6 (2.3) 11.5 (3.0) 5.3 (1.4) 11.5 (5.8) 5.2 (1.5)

.039 .019 .008 NS (.406) .003 NS (.115) .016 .040

NOTE. Values are median survival times in months with standard errors in parentheses.

HEPATOLOGY, Vol. 35, No. 5, 2002

LO ET AL.

1169

Table 5. Comparison of Liver Function as Assessed by the Serum Bilirubin Level, Serum Albumin Level, and Indocyanine Green Retention Rate at 15 Minutes
Chemoembolization Group No. of Patients Median (Interquartile Range) No. of Patients Control Group Median (Interquartile Range)

Bilirubin ( mol/L) 3 Months 6 Months 9 Months 12 Months Albumin (g/L) 3 Months 6 Months 9 Months 12 Months Indocyanine green retention at 15 minutes (%) 3 Months 6 Months 9 Months 12 Months

34 24 19 17 34 24 19 17 32 25 22 17

15 (10-31) 14 (11-24) 16 (10-20) 13 (9-23) 36 (31-40) 35 (31-39) 35 (31-37) 33 (31-39) 25 (14-43) 25 (18-38) 25 (20-34) 26 (13-34)

24 13 7 9 24 13 7 9 25 13 10 9

21 (14-42) 17 (10-28) 18 (12-32) 15 (11-30) 32 (26-37) 35 (28-38) 33 (28-36) 34 (27-39) 36 (20-52) 26 (17-51) 28 (15-48) 32 (22-47)

.038 .987 .385 .517 .073 .425 .223 .499 .169 .433 .405 .146

transarterial Lipiodol chemoembolization with the present regimen induced a marked tumor response and signicantly prolonged the survival of this selected group of Asian patients with unresectable hepatocellular carcinoma and preserved liver functions. By adjusting for baseline prognostic factors in a Cox model, treatment with chemoembolization remained a signicant independent factor that reduced the hazard of death by half. The differences in patient population and methodology account for the differences between the positive result of our trial as compared with the absence of survival benet reported in previous randomized controlled trials.17-19 One trial from South Africa18 mostly included black or mixed-race patients with advanced tumors, and the conclusion would not be valid for other patients. The other 2 randomized trials from France17,19 included European patients, mainly with alcohol-induced liver disease. Despite obvious evidence of tumor response in both trials, the survival benet might have been offset by the adverse effects of chemoembolization because of the poor tolerance of patients with alcoholic cirrhosis to this form of treatment.19 In addition, there are ethnic differences in the behavior of hepatocellular carcinoma20 and the natural history of untreated disease in European patients27,28 is much better than that of Asians.24 It has been shown that untreated patients who had positive results for hepatitis B surface antigen had a worse prognosis.28 Our study included Asian patients with mostly hepatitis Brelated hepatocellular carcinoma of Okuda stage I or II. The 2-year survival rate of 11% in our control group was comparable with our estimation based on the published data in Asian patients24 and was lower than the 26% 2-year

survival rate of untreated patients as reported in the French multicenter trial.19 The technique and regimen of chemoembolization also affect the outcome of treatment. In our study, we used cisplatin-Lipiodol emulsion and gelatin sponge particles. This is the most generally accepted method for chemoembolization. However, our technique and regimen differed from those of the previous trials in several ways. First, to maximize the efcacy and to minimize toxicity, chemoembolization was given by selective injection into the feeding artery in nearly half of our cases. Second, we used a lower dosage of cisplatin in our trial (median of 10 mg) compared with that of the French multicenter trial19 (xed dosage of 70 mg), and this may account for the patients better tolerance of treatment. Third, in contrast to the xed dosage of antitumor agent that was used in the other studies,17-19 we used a variable dosage of cisplatin-Lipiodol emulsion that was based on the tumor size.11,13,14 As chemoembolization is a form of targeted therapy, the amount of drug required should be tailored according to the size of the target to minimize the adverse effects caused by the excessive amount of drug delivered to the normal liver tissue. Finally, we did not set any limit to the number of treatment courses in our trial. Provided the patient tolerated the procedure and the tumor continued to respond, chemoembolization was repeated at regular intervals to continue the suppression of tumor growth. Half of the patients received more than 4 courses of treatment and as many as 15 courses were given to 1 patient. Using such a regimen, objective response rate in serum -fetoprotein level was 72% and in tumor size 39%. The

1170

LO ET AL.

HEPATOLOGY, May 2002

lower response rate in tumor size may be attributed to the persistent Lipiodol staining in areas of the tumor that might not contain viable cancer cells. This benet of tumor suppression was not offset by the adverse effect on the function of the normal liver. Although there were more deaths attributed to liver failure in patients who received chemoembolization, this might reect partly the natural course of the underlying liver disease when death caused by tumor progression has been effectively delayed with chemoembolization. On the other hand, for untreated patients with advanced tumor, tumor progression was recognized as the cause of death even though features of liver failure were usually present. In any case, the adverse effect on liver function was not observed in the survivors. With a signicantly longer survival in patients who received chemoembolization, the liver function of survivors was at least no worse than that of the control group. In addition, there was no signicant increase in the need for hospital admission for adverse events. The 2-year survival rate was increased to 31% and this was consistent with our previous experience.13 The prognosis of hepatocellular carcinoma is correlated with factors related to the extent of tumor and hepatic function.24,29-31 In our study, all the prognostic factors for survival were related to tumor growth and none was related to liver function, because patients with decompensated cirrhosis were excluded. To further improve the results of Lipiodol chemoembolization, we stratied the patients according to the baseline prognostic variables. The survival benet of treatment was maintained in all the subgroups except those with large tumors or unilobal portal vein obstruction. However, it is important to note that the comparison of survival between the treatment and control group after stratication may result in a loss of statistical power. Chemoembolization improved the median survival of patients with tumors larger than 5 cm from 5.3 months to 11.2 months, although the difference was not statistically signicant (P .115). There were only 21 patients with unilobar portal vein obstruction, and it would be difcult to show a signicant survival benet with chemoembolization in this subgroup of patients. Nevertheless, the presence of unilobar portal vein obstruction has remained with chemoembolization treatment as the 2 independent prognostic factors in the multivariate Cox model. The presence of unilobar portal vein obstruction increased the risk of death by almost 3 times and the median survival with and without chemoembolization was only 5.1 months and 2.6 months, respectively. Because the dismal prognosis is not signicantly improved by chemoembolization, it may not be appropriate to recommend this form of treatment for this subgroup of patients.

In conclusion, transarterial Lipiodol chemoembolization using the present regimen prolongs the survival of a selected group of Asian patients with unresectable hepatocellular carcinoma and is an effective palliative treatment option. Whether non-Asian patients with this disease condition will benet from a similar regimen or other regimens of chemoembolization remains to be determined by further randomized controlled trials. Acknowledgment: The authors thank Dr. To-Wai Leung of the Department of Clinical Oncology, Tuen Mun Hospital, for his constructive comments for the manuscript.

References
1. Xavier Bosch F: Global epidemiology of hepatocellular carcinoma. In: Okuda K, Tabor E, eds. Liver Cancer. New York: Churchill Livingstone, 1997;13-28. 2. Bruix J. Treatment of hepatocellular carcinoma. HEPATOLOGY 1997;25:259-262. 3. Mor E, Kaspa RT, Sheiner P, Schwartz M. Treatment of hepatocellular carcinoma associated with cirrhosis in the era of liver transplantation. Ann Intern Med 1998;129:643-653. 4. Sasaki Y, Imaoka S, Kasugai H, Fujita M, Kawamoto S, Ishiguro S, Kojima J, et al. A new approach to chemoembolization therapy for hepatoma using ethiodized oil, cisplatin, and gelatin sponge. Cancer 1987;60:1194-1203. 5. Kajanti M, Rissanen P, Virkkunen P, Franssila K, Mantyla M. Regional intra-arterial infusion of cisplatin in primary hepatocellular carcinoma. A phase II study. Cancer 1986;58:2386-2388. 6. Kanematsu T, Furuta T, Takenaka K, Matsumata T, Yoshida Y, Nishizaki T, Hasuo K, et al. A 5-year experience of lipiodolisation: selective regional chemotherapy for 200 patients with hepatocellular carcinoma. HEPATOLOGY 1989;10:98-102. 7. Wheeler PG, Melia W, Dubbins P, Jones B, Nunnerley H, Johnson P, Williams R. Non-operative arterial embolization in primary liver tumors. Br Med J 1979;2:242-244. 8. Lin DY, Liaw YF, Lee TY, Lai CM. Hepatic arterial embolization in patients with unresectable hepatocellular carcinomaa randomized controlled trial. Gastroenterology 1988;94:453-456. 9. Bruix J, Llovet JM, Castells A, Montana X, Bru C, Ayuso MDC, Vilana R, et al. Transarterial embolization versus symptomatic treatment in patients with advanced hepatocellular carcinoma: results of a randomized, controlled trial in a single institution. HEPATOLOGY 1998;27:1578-1583. 10. Kasugai H, Kojima J, Tatsuta M, Okuda S, Sasaki Y, Imaoka S, Fujita M, et al. Treatment of hepatocellular carcinoma by transcatheter arterial embolization combined with intraarterial infusion of a mixture of cisplatin and ethiodized oil. Gastroenterology 1989;97:965-971. 11. Vetter D, Wenger JJ, Bergier JM, Doffoel M, Bockel R. Transcatheter oily chemoembolization in the management of advanced hepatocellular carcinoma in cirrhosis: results of a Western comparative study in 60 patients. HEPATOLOGY 1991;13:427-433. 12. Bismuth H, Morino M, Sherlock D, Castaing D, Miglietta C, Cauquil P, Roche A. Primary treatment of hepatocellular carcinoma by arterial chemoembolization. Am J Surg 1992;163:387394.

HEPATOLOGY, Vol. 35, No. 5, 2002

LO ET AL.

1171

13. Ngan H, Lai CL, Fan ST, Lai EC, Yuen WK, Tso WK. Treatment of inoperable hepatocellular carcinoma by transcatheter arterial chemoembolization using an emulsion of cisplatin in iodized oil and gelfoam. Clin Radiol 1993;47:315-320. 14. Stefanini GF, Amorati P, Biselli M, Mucci F, Celi A, Arienti V, Roversi R, et al. Efcacy of transarterial targeted treatments on survival of patients with hepatocellular carcinoma. Cancer 1995; 75:2427-2434. 15. Bayraktar Y, Balkanci F, Kayhan B, Uzunalimoglu B, Gokoz A, Ozisik Y, Gurakar A, et al. A comparison of chemoembolization with conventional chemotherapy and symptomatic treatment in cirrhotic patients with hepatocellular carcinoma. Hepato-Gastroenterology 1996;43:681-687. 16. Stuart K, Stokes K, Jenkins R, Trey C, Clouse M. Treatment of hepatocellular carcinoma using doxorubicin/ethiodized oil/gelatin powder chemoembolization. Cancer 1993;72:3202-3209. 17. Pelletier G, Roche A, Ink O, Anciaux ML, Derhy S, Rougier P, Lenoir C, et al. A randomized trial of hepatic arterial chemoembolization in patients with unresectable hepatocellular carcinoma. J Hepatol 1990;1:181-184. 18. Madden MV, Krige EJ, Bailey S, Beningeld SJ, Geddes C, Werner ID, Terblanche J. Randomised trial of targeted chemotherapy with lipiodol and 5-epidoxorubicin compared with symptomatic treatment for hepatoma. Gut 1993;34:1598-1600. 19. Groupe dEtude et de Traitement du Carcinome Hepatocellulaire. A comparison of lipiodol chemoembolization and conservative treatment for unresectable hepatocellular carcinoma. N Engl J Med 1995;332:1256-1261. 20. Chin PL, Chu DZJ, Clarke KG, Odom-Maryon T, Yen Y, Wagman LD. Ethnic differences in the behavior of hepatocellular carcinoma. Cancer 1999;85:1931-1936. 21. Beasley RP, Hwang LY, Lin CC, Chien CS. Hepatocellular carcinoma and hepatitis B virus. A prospective study of 22,707 men in Taiwan. Lancet 1981;2:1129-1133.

22. Falkson G, Maclntyre JM, Moertel CG, Johnson LA, Scherman RC. Primary liver cancer. An Eastern Cooperative Oncology Group trial. Cancer 1984;54:970-977. 23. Kawasaki S, Sugiyama Y, Iga T, Hanano M, Sanjo K, Beppu T, Idezuki Y. Pharmacokinetic study on the hepatic uptake of indocyanine green in cirrhotic patients. Am J Gastroenterol 1985;80: 801-806. 24. Okuda K, Ohtsuki T, Obata H, Tomimatsu M, Okazaki N, Hasegawa H, Nakajima Y, et al. Natural history of hepatocellular carcinoma and prognosis in relation to treatment: study of 850 patients. Cancer 1985;56:918-928. 25. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1968;53:457-481. 26. Cox DR. Regression models and life-tables. J Roy Stat Soc 1972; 34:187-220. 27. Llovet JM, Bustamante J, Castells A, Vilana R, Ayuso MD, Sala M, Bru C, et al. Natural history of untreated nonsurgical hepatocellular carcinoma: rationale for the design and evaluation of therapeutic trials. HEPATOLOGY 1999;29:62-67. 28. Villa E, Moles A, Ferretti I, Buttafoco P, Grottola A, Del Buono M, De Santis M, et al. Natural history of inoperable hepatocellular carcinoma: estrogen receptors status in the tumor is the strongest prognostic factor for survival. HEPATOLOGY 2000;32:233238. 29. Rosellini SR, Arienti V, Nanni O, Ugenti F, Tassinari M, Camporesi C, Boriani L, et al. Hepatocellular carcinoma. Prognostic factors and survival analysis in 135 Italian patients. J Hepatol 1992;16:66-72. 30. Nomura F, Ohnishi K, Tanabe Y. Clinical features and prognosis of hepatocellular carcinoma with reference to serum -fetoprotein levels. Analysis of 606 patients. Cancer 1989;64:1700-1707. 31. Calvet X, Bruix J, Gines P, Bru C, Sole M, Vilana R, Rodes J. Prognostic factors of hepatocellular carcinoma in the west: a multivariate analysis in 206 patients. HEPATOLOGY 1990;12:753-760.