Trans-catheter arterial chemoembolisation for hepatocellular carcinoma in patients with viral cirrhosis: role of combined staging systems, Cancer Liver Italian Program (CLIP) and Model for End-stage Liver Disease (MELD), in predicting outcome after treatment


Prof Roberto Testa, Gastroenterology Unit, Department of Internal Medicine (DI.M.I), Viale Benedetto XV n°6, 16132 Genova, Italy.


Background:  Trans-catheter arterial chemoembolisation (TACE) is the most common palliative treatment for hepatocellular carcinoma (HCC). The therapeutic options depend both on the characteristics of the tumour and on functional staging of the cirrhosis.

Aim:  To evaluate the effects of TACE on the survival of cirrhotic patients with HCC according to different staging systems [Okuda score, Cancer Liver Italian Program (CLIP) score, Model for End-stage Liver Disease (MELD) score] and in relation to the side-effects of TACE.

Methods:  Fifty cirrhotic patients, 36 CTP class A and 14 class B, underwent 106 TACE treatments with mitoxantrone. Survival at 12, 24, and 36 months was evaluated.

Results:  MELD at 12 months and CLIP at 24 months were identified as significant variables associated with survival. Combined cut-offs of CLIP and of MELD identified four subgroups of patients with different survivals, at 12, 24 and 36 months, respectively: CLIP ≥ 2 and MELD ≥ 10 (63%, 20% and 0%), CLIP < 2 and MELD ≥ 10 (73%, 40% and 22%), CLIP ≥ 2 and MELD < 10 (73%, 40% and 22%) and CLIP < 2 and MELD < 10 (100%, 63% and 50%). Post-TACE side-effects proved to have no influence on survival.

Conclusion:  In patients with poor probability of survival (CLIP ≥ 2 and MELD ≥ 10), TACE must be planned with a great deal of caution, while in patients with possibly good outcomes (CLIP < 2 and MELD < 10), more ‘aggressive’ therapy should be taken into consideration.


In Western countries, hepatocellular carcinoma (HCC) is commonly associated with cirrhosis.1 Decisions regarding therapy for HCC depend on the number, size and location of the tumour(s) in the liver, as well as on the functional staging of cirrhosis.2–4 At diagnosis few patients can benefit from curative treatment (surgery or locoregional ablation).2 Trans-catheter arterial chemoembolisation (TACE) is the palliative therapeutic option for many patients who are not suitable to undergo curative treatment.2, 5, 6 TACE can be employed in patients with preserved liver function, with no massive tumour and without portal thrombosis.5, 6 Standard staging of cirrhosis is the Child–Turcotte–Pugh score (CTP), and CTP C patients are usually excluded from TACE.6 A staging system that takes into consideration both the liver tumour and liver function is the Okuda score, and more recently the Cancer of the Liver Italian Program (CLIP) score has been proposed.7–9 Furthermore, MELD (Model for End-stage Liver Disease), a novel, functional system for staging cirrhosis, has also recently been validated.10–12 This prognostic scoring method has been applied to cirrhotic patients not affected with HCC in a transplant perspective. It has also been used to evaluate the risk of transjugular intrahepatic portosystemic shunt (TIPS) worsening liver function.10 Many papers have pointed out that TACE decreases tumour progression, but does not improve survival.13–15 In 2002, two papers based on studies in Asia and Europe showed that TACE improves survival, and therefore can be considered effective treatment in selected patients.16, 17 TACE per se is a procedure that can affect hepatic function,5, 6 thus the aim of our study was to evaluate the survival of cirrhotic patients with HCC, both in relation to different staging systems and in relation to TACE side-effects on the liver.

Patients and methods

All cirrhotic patients with unresectable hepatocellular carcinoma (HCC) presenting at our Unit between September 1994 and September 2001 were reviewed. We identified 142 patients for whom therapeutic intervention was chosen on the basis of the clinical and functional staging of the liver disease and on the characteristics of the tumour. Fifty-six patients had undergone TACE as palliative treatment, and informed consent was obtained from all patients. This study included 50 patients with virus-related cirrhosis, with a follow-up of almost 12 months, and for whom complete clinical and biochemical records were available. The group was made up of 14 females and 36 males, aged 48–81 years, 36 CTP class A and 14 class B. Cirrhosis was related to hepatitis C virus in 38 patients and to hepatitis B virus in five, and an HCV–HBV coinfection was present in seven patients. Diagnosis of HCC was confirmed by fine-needle biopsy in 16 patients, by an unequivocal radiological finding associated with serum α-fetoprotein greater than 400 ng/mL in nine patients, and by two coincidental imaging techniques associated with increased α-fetoprotein in 25 patients.2 Thirty-two patients had multifocal HCC and 18 had unifocal HCC. On the basis of the pertinent parameters, Okuda score, CLIP score, and MELD score were calculated at the time of the first TACE, according to original papers.7, 10 Briefly CLIP criteria are: CTP stage (A, B and C; score 0–2), tumour morphology (uninodular, multinodular, massive or extension > 50% of tumour; score 0–2), α-fetoprotein levels (< 400 ng/mL, > 400 ng/mL; score 0–1), macrovascular invasion (no, yes; score 0–1). The MELD score was calculated according to the original formula proposed by the Mayo Clinic group: 3.8 × ln[bilirubin (mg/dL)] + 11.2 × ln(INR) + 9.6 × ln[creatinine (mg/dL)] + 6.4 × (aetiology: 0 if cholestatic or alcoholic, 1 otherwise). Exclusion criteria from TACE were: age above 75 years (four patients were exonerated from this limit owing to very good performance status), advanced liver disease (CTP class C), presence of vascular invasion (including segmental portal obstruction), extrahepatic spread, any contraindication to an arterial procedure such as impaired clotting tests, renal failure, severe atheromatosis (as per TAC findings), and any contraindications to mitoxantrone. All patients underwent TACE following a standard protocol. Intravenous ceftriaxone (1 g b.d.) was given from the night before to 3 days after the procedure as antibiotic prophylaxis. The patients had fasted overnight, and intravenous ondasetron (10 mg) and ranitidine (100 mg) were administered just before the procedure. TACE was performed by expert radiologists (G.D.C., G.C.), using mitoxantrone (20 mg standard dose, reduced by 5% for each 0.1 mg of bilirubin exceeding the limit of 2 mg/dL6) and lipiodol (10 mL) before mechanical obstruction. After local anaesthesia the femoral artery was percutaneously catheterised, and the catheter was inserted into the main hepatic artery under arteriography in order to define tumour nodules. Whenever possible, the hepatic artery feeding the tumour was super-selectively catheterised. All procedures were performed under radiographic control, and a control angiography was performed immediately after embolisation. All post-TACE clinical side-effects were recorded, and biochemical and haematological data were monitored for at least 3 days. Treatment response was assessed by contrast-enhanced spiral computed tomography at either 3 or 4 weeks following treatment. Patients were assessed every 2 months by means of echotomography, and TACE was repeated on ‘demand’. Nineteen patients underwent one TACE treatment, 17 underwent two, six underwent three, five underwent four, and three patients underwent five. Survival was calculated in months from the time of the first TACE. Follow-up ranged from 12 to 72 months.

Statistical analysis

Results are expressed as mean ± standard deviation (s.d.) or median (range) and 95% confidence limits (CL). Comparisons between groups were performed using Fisher's exact test and the Mann–Whitney U-test. Univariate survival curves were estimated using the Kaplan–Meier method. The χ2 test for combined score evaluation was performed in sequence, and sensitivity, specificity and negative and positive predictive values were calculated for the best cut-offs, identified by means of the ROC curves. The relationships among post-treatment δ% of the biochemical variables were evaluated by means of Spearman's rank correlation coefficient. A P-value < 0.05 for two-sided tests was considered statistically significant. Statistical analysis was performed using the SPSS statistical package (SPSS, Chicago, IL), and ROC curves were calculated using MedCalc 5.00 statistical software (MedCalc Software, Mariakerke, Belgium).


Fifty cirrhotic patients underwent 106 TACE treatments for unresectable HCC. Survival at 6, 12, 24 and 36 months was 94% (47/50), 80% (40/50), 39% (16/41) and 16% (5/32) (95% CL was, respectively, 84–99, 66–90, 24–55 and 5–33). Table 1 reports the pre-TACE staging system scores on the basis of 12 and 24 months of survival. Significant differences between deceased and surviving patients have only been found for MELD at 12 months and for CLIP at 24 months. Table 2 shows percentage survival at 12 and 24 months according to combined cut-offs of MELD and of CLIP. Among the combinations above and below both cut-offs, survival at 12 months ranged between 100% and 63%, and between 63% and 20% at 24 months. The performance of the combined cut-offs, identified by means of ROC curves (data not shown), of MELD (≥ 10) and of CLIP (≥ 2) showed a specificity of 70% (95% CL 53.5–83.4) and a negative predictive value of 90.3 (95% CL 74.2–98.0) at 12 months, and a specificity of 81.3% (95% CL 54.4–96.0) and positive predictive value of 80% (95% CL 51.9–95.7) at 24 months. Survival at 36 months according to the sequence (as in Table 2) of the combined cut-offs of CLIP and MELD were, respectively, 0% ( CL 0–26), 22%, (CL 3–60), 25% (CL 3–65) and 50% (CL 12–88). The specificity of the combined cut-offs was 100% (CL 63–100) together with a positive predictive value of 100% (CL 73.5–100). On the basis of these predictive cut-offs (Figure 1), the probable benefit of TACE for cirrhotic patients whose MELD is below 10 and whose CLIP is below 2 could be over 90% at 12 months, over 50% at 24 months and 40% at 36 months. The patients whose MELD is ≥ 10 and whose CLIP is ≥ 2 would have a probable TACE benefit of 40% at 12 months, and no prospective of survival at 24 months. Table 3 summarises the post-TACE clinical and biochemical side-effects: fever was the most common finding, and prolonged hospitalisation in patients since it persisted for more than 3 days. Liver failure also prolonged hospitalisation in some patients. Myelotoxic effects were only observed in some patients and were reversible. Increases greater than 25% of the baseline values were observed for bilirubin in 40% of TACE procedures, for aspartate aminotransferase (AST) in 44%, for alanine aminotransferase (ALT) in 39%, for lactic dehydrogenase (LDH) in 25% and for γ-glutamyl transpeptidase (GGT) in 8%(Table 4). Among liver function tests, only δ (% bilirubin) showed no significant correlations with δ (% enzymatic tests), which however, showed significant inter-relations between AST and ALT (r = 0.890, P = 0.0001), AST and GGT (r = 0.402, P =0.001), AST and LDH (r = 0.522, P = 0.001), AST and APH (r = 0.308, P = 0.01), ALT and GGT (r = 0.257, P = 0.01), ALT and LDH (r = 0.510, P = 0.001), ALT and APH (r = 0.321, P = 0.01), GGT and APH (r = 0.368, P = 0.01). Patients who showed either an increase > 50% in bilirubin serum levels or both an increase > 25% but < 50% in bilirubin serum levels and an increase > 25% of at least three enzymatic tests were considered patients with post-TACE impairment of liver function. Significant differences in pre-treatment variables among patients with and without impaired liver function were observed for platelets, white blood cells, alkaline phosphatase, lactic dehydrogenase and bilirubin (Table 5). All significant parameters were more deranged in patients who did not experience hepatic side-effects from TACE. The presence of impaired liver function did not modify the survival of the patients: impaired vs. non-impaired, 22 ± 20 vs. 20 ± 13 months (P = 0.789).

Table 1.  Pre-TACE clinical staging scores according to survival at 12 and 24 months
Score12 months P24 months P
Deceased n =10Survivors n =40Deceased n= 25Survivors n = 16
CTP7.0 ± 2.25.9 ± ± 1.85.6 ± 0.90.14
OKUDA1.0 ± 1.00.5 ± ± 0.80.3 ± 0.50.074
CLIP2.2 ± 1.31.4 ± ± 1.21.1 ± 1.00.038
MELD13.2 ± 4.39.8 ± 1.30.03511.0 ± 4.39.4 ± 2.60.18
Table 2.  Outcome at 12 and 24 months according to combined staging cut-off of CLIP and MELD scores
CLIP scoreMELD score12 months24 months
Deceased % 95% CLSurvivors % 95% CLDeceased % 95% CLSurvivors % 95% CL
≥  2 ≥  10 37% (16–67)  63% (38–84) 80% (52–96) 20% (4–48)
< 2 ≥  10 27% (6–61)  73% (39–94) 60% (26–88) 40% (12–74)
≥  2< 10  0% (0–37) 100% (63–100) 50% (16–84) 50% (16–84)
< 2< 10  0% (0–26) 100% (74–100) 37% (9–76) 63% (25–92)
Figure 1.

Kaplan–Meier estimated survival curves at 36 months by combined CLIP and MELD scores.

Table 3.  Clinical complications of 106 TACE treatments
  • *

     Combined platelet and white blood cell decrease > 25%, 4/106.

  • Isolated platelet decrease > 25%, 18/106.

  • Isolated white blood cell decrease > 25%, 2/106.

Bleeding/Bruising haematoma  8/106
Gastrointestinal bleeding  2/106
Development of ascites  5/106
Hepatic encephalopathy  1/106
Table 4.  Impaired liver function tests after each TACE procedure
 Minor impaired liver tests Increase in baseline value > 25% < 50%Major impaired liver tests Increase in baseline value > 50%
Event/TACE%95% ClEvent/TACE%95% Cl
  1. BIL, bilirubin; AST, aspartate aminotransferase; ALT, alanine aminotransferase; GGT, γ-glutamyl transpeptidase; LDH, lactic dehydrogenase.

BIL (mg/dL)16/10614%8–2327/10626%17–36
AST (IU/mL)11/10610%5–1736/10634%25–45
ALT (IU/mL)  9/106  9%4–1631/10630%21–40
LDH (IU/mL)11/10610%4–1716/10615%8–24
GGT (IU/mL)  6/106  5%2–11  3/106  3%0–6
Table 5.  Overall pre-treatment values before each TACE procedure according to development of liver function impairment post-TACE (preserved or impaired LF)
 Post-TACE preserved LF N = 62Post-TACE impaired LF N = 44P
  1. AST, aspartate aminotransferase; ALT, alanine aminotransferase; GGT, γ-glutamyl transpeptidase; APH, alkaline phosphatase; LDH, lactic dehydrogenase; BIL, bilirubin; PA, prothrombin activity.

Hb (g/dL)13.0 ± 2.013.7 ± 1.50.1026
PLT (/mm3)97 ± 55121 ± 660.0302
WBC (/mm3)4378 ± 14845352 ± 19600.0083
AST (IU/mL)96 ± 4796 ± 670.487
ALT (IU/mL)93 ± 7093 ± 540.627
GGT (IU/mL)94 ± 95103 ± 770.993
APH (IU/mL)327 ± 140263 ± 990.0153
LDH (IU/mL)412 ± 89376 ± 1030.0039
BIL (mg/dL)1.5 ± 0.81.0 ± 0.40.0003
PA (%)53 ± 3858 ± 400.2244


The key problem in the treatment of HCC is the associated cirrhosis, which limits therapeutic choices. As far as palliative therapy is concerned, an increase in survival with no impairment of the quality of life should be the goal.18 Recently, a significant benefit in survival has also been found for TACE.16, 17, 19, 20 We have analysed traditional and novel staging systems as possible predictors of survival, and of post-TACE liver function impairment. This study is open and not randomised. It was planned on the basis of intention to treat all cirrhotic patients with HCC for whom there was an indication to TACE.6 The percentage survival observed in our patients proved to be in line with RCTs reported in recent reviews13–15, 19 both at 12 and 24 months. The bases for a rough comparison are the conventional criteria to perform TACE in cirrhotic patients with HCC.6 Furthermore, this also allowed us to compare survival against that of patients who had only been treated conservatively. Tumour response after TACE was not considered a predictor of survival owing to two problems: first because the criteria to assess tumour response properly are not standardised;6 secondly, our policy with regard to performing further TACE ‘on demand’ during close follow-up limits the role played by the tumour response in the analysis of survival. Both randomised and non-randomised trials have shown that overall survival for TACE ranges from 9 to 21 months.15 The median survival of 102 cirrhotic patients with unresectable HCC who were managed with symptomatic treatment was 17 months, and the overall actuarial probability of survival at 1, 2 and 3 years was 54%, 40% and 28%, respectively.21 Two recent papers have assessed the effectiveness of TACE in survival.16, 17 Tumour-related symptoms were found in 25% of TACE patients and in 51% of controls in the paper by Llovet et al.,17 and in 60% of the cases studied by Lo et al.16 Llovet et al. showed a 1- and 2-year survival probability of 82% and 63%, respectively, for the chemoembolisation group, and 63% and 27% for the control group, while Lo et al. showed significant survival at minor levels both in the chemoembolisation group (57% and 31%) and in the control group (32% and 11%). This apparent discrepancy likely depends on a prevalence of more advanced stages of disease in the Far-Eastern patients, and points out that survival is significant when both the treated and the control groups are comparable for staging. Therefore, both papers describe two subgroups which benefited from TACE. The papers identified two adverse risk factors which exist in the absence of treatment options and in the presence of symptoms; Llovet et al. associated bilirubin and Lo et al. associated tumour parameters. Overall survival in our patients was in keeping with Llovet et al. at 12 months and with Lo et al. at 24 months, suggesting that global staging of our patients was intermediate with respect to the two others, as pointed out by both Okuda and Child–Turcotte–Pugh stage distribution among the groups. In our patients, neither extra-hepatic spread nor portal vascular invasion was present, but constitutional syndrome was present in over half of the patients. In our patients, mid-term survival proved to be related to the staging of cirrhosis, as shown by MELD, while in the long term, survival depends both on the tumour and on the cirrhosis status, as assessed by CLIP. The role that CLIP plays is in keeping with the prognostic index that separately identifies α-fetoprotein, tumour size and CTP score as prognostic predictors for TACE.22 Furthermore, our data confirm a previous study involving patients undergoing TACE in which the CLIP score was selected as the most significant single independent predictor in the Cox regression.23 No previous studies have tested MELD as a prognostic indicator for cirrhosis with HCC, particularly in evaluating the outcome after treatment, but all have taken CTP score into consideration. In evaluating the outcome of TACE, the role of the CTP score has probably been a confounding factor owing to its use as a discriminant probe for performing this treatment (CTP C patients are excluded from treatment). This confounding factor is also present both in the Okuda score and in the CLIP score, which consider CTP as part of their respective scores. The MELD score seems to overcome the limited discriminant role that CTP score11 plays within CLIP. As a matter of fact, when combined, MELD and CLIP seem to play a further role: by subdividing patients on the basis of the four combinations, we extrapolated a group that obtained apparently high benefit from TACE, and a group apparently achieving very little benefit from TACE. Two other intermediate groups showed a possible benefit, which is comparable to the results of Lo et al. It is noteworthy that the range of survival for each sub-group that was identified by means of the CLIP–MELD combination was in keeping with the large range reported at 1, 2 and 3 years in the reviews both by Alsowmely and Hodgson15 and by Llovet and Bruix.20 In our patients both the MELD and CLIP staging systems co-operate towards pre-TACE prognosis; furthermore, MELD seems to be the main determinant of the outcome of patients undergoing TACE therapy. The role that MELD plays in predicting survival of HCC patients with liver cirrhosis can be further emphasised by the fact that many deaths are due to cirrhosis and not to hepatocellullar carcinoma.24 Nevertheless, it is noteworthy that two variables of the MELD score (bilirubin and INR) are included in the CTP score, which is part of CLIP. MELD likely identified other subgroups with different survivals among cirrhotic patients in CTP classes A or B.25

The side-effects of TACE are important with regard to quality of life. We observed some myelotoxic effects, likely related to mitoxantrone, and which were comparable to similar effects using doxorubicin or cisplatin.22, 23, 26, 27 Few cases of cirrhosis decompensation were observed, likely due to the fact that TACE was prevalently performed selectively. However, frequent liver function impairment was observed, although it was always reversible, and comparable in entity and frequency to other reports.27 Bilirubin can be considered an independent parameter both owing to the predictive power of survival for TACE22, 23 and to the fact that in our series its modifications are not related to enzymatic modifications. Therefore, both bilirubin and enzymatic increases were considered separate findings of liver dysfunction. Surprisingly, the patients who experienced greater liver function impairment had lower clinical staging of cirrhosis (MELD 9.2 ± 3.2 vs. 11.9 ± 3.5, P = 0.0042). These side-effects do not seem to be related to the mitoxantrone dose. Both major and minor impairment of liver function was observed in patients who were administered reduced doses on the basis of their bilirubin levels. It would appear that the more stable patients are more susceptible to ischaemic damage, and monitoring hepatic function by means of a quantitative liver function test could probably help clarify this observations and evaluate the risk of hepatic failure after TACE more precisely.28 Impairment of hepatic function does not affect survival, but it certainly worsens the quality of life of these patients. Therefore, this must be taken into consideration when TACE is planned in cirrhotic patients with HCC.29 The complications related to TACE apparently do not affect survival, and considering that these side-effects are unpredictable by means of pre-treatment biochemical status, TACE could be offered to all patients not presenting any contraindications. Nevertheless, with regard to the risk–benefit decision, the survival prospective of the single patient must be taken into consideration on the basis of combined staging (MELD and CLIP), of age, which is frequently advanced when HCC is diagnosed,30 and of the possible morbidity of side-effects.19 In conclusion, in our study we observed that among the HCC patients with TACE indication, those with poor combined staging of MELD and CLIP must be evaluated carefully and, in a in strict risk–benefit evaluation, should not be taken into consideration for TACE. Conversely, more aggressive treatment, such as TACE plus radiofrequency ablation, could be planned for patients for whom both staging systems predict a favourable outcome.31–34 These suggestions must be validated in a larger population and in several clinical centres.