Treatment of small hepatocellular carcinoma with percutaneous ethanol injection: a validated prognostic model

THE AMERICAN JOURNAL OF GASTROENTEROLOGY © 2000 by Am. Coll. of Gastroenterology Published by Elsevier Science Inc.

Vol. 95, No. 10, 2000 ISSN 0002-9270/00/$20.00 PII S0002-9270(00)01998-5

Treatment of Small Hepatocellular Carcinoma With Percutaneous Ethanol Injection: A Validated Prognostic Model Ambrogio Orlando, M.D., Adele D’Antoni, M.D., Calogero Camma`, M.D., Maddalena Albanese, M.D., Tito Livraghi, M.D., Guido Torzilli, M.D., Roberto Virdone, M.D., Elio Sciarrino, M.D., Rosa Giovanna Simonetti, M.D., Alberto Maringhini, M.D., Luigi Pagliaro, M.D., and Mario Cottone, M.D. Divisione di Medicina Interna, Ospedale V. Cervello, Palermo, Italy; Istituto di Medicina Generale e Pneumologia, University of Palermo, Palermo, Italy; Istituto Metodologie Diagnostiche Avanzate, Consiglio Nazionale delle Ricerche, Palermo, Italy; Servizio Radiologia Ospedale Civile, Vimercate, Italy; and Divisione di Chirurgia Generale Ospedale Civile, Lodi, Italy

OBJECTIVE: Percutaneous ethanol injection may prolong the survival of patients with small hepatocellular carcinoma associated with cirrhosis. The aim was to identify prognostic factors of survival and of local recurrence, as well as separate new lesions. METHODS: We performed Cox regression analysis in 115 consecutive patients with hepatocellular carcinoma (81 Child-Pugh class A, 34 Child-Pugh class B) treated by percutaneous ethanol injection. The validity of the model was tested by comparing predicted and observed survival in 105 independent patients from an external series. RESULTS: Overall survival rates were 89%, 63%, and 43% at 1, 2, and 3 yr, respectively. The 1-, 2-, and 3-yr survival rates were 96%, 78%, and 63%, respectively, for ChildPugh class A patients and were 73%, 35%, 12%, respectively, for Child-Pugh class B. The albumin level was the only independent variable significantly associated with survival (p ⬍ 0.0001). The 3-yr rate of appearance of separate new lesions and local recurrence were 41% and 23%, respectively. The survival predicted by the model agreed with that observed in the independent patients. CONCLUSIONS: Survival of patients with hepatocellular carcinoma treated by percutaneous ethanol injection is related to baseline albumin level. The high rate of recurrence (both local and distant) points out the palliative role of this therapy. (Am J Gastroenterol 2000;95:2921–2927. © 2000 by Am. Coll. of Gastroenterology)

INTRODUCTION Percutaneous ethanol injection (PEI) may prolong the survival of patients with hepatocellular carcinoma (HCC) associated with cirrhosis, as suggested by nonrandomized trials with historical untreated controls (1, 2) and by a cohort study comparing surgical resection with PEI (3). Several studies have been carried out to identify prognostic factors

for response to PEI (4 –7). Only the study of Castellano et al. (7) used a multivariable model identifying serum albumin, number of lesions, and ␣-fetoprotein (AFP) levels as independent variables associated with survival. However, the limitations of this paper are the imbalance between number of events and number of variables included in the model and the lack of validation. We report the results of a prospective study carried out to identify the type of patient who might benefit most from PEI in a cohort of patients with small HCC (ⱕ5 cm) associated with cirrhosis, using a multivariable analysis and validating the model in an external set of patients. Our secondary aims were to assess the risk factors for separate new lesions and local recurrence.

MATERIALS AND METHODS From February 1992 to June 1997, 115 white patients with ultrasound (US)– guided, biopsy-proven HCC (ⱕ5 cm) associated with cirrhosis were treated by PEI. This cohort of patients was selected from a population of 207 consecutive subjects admitted to our institution for HCC. We used the following criteria to select the patients for PEI: age ⬎60 yr, or age ⬍60 yr in patients who refused surgery or orthotopic liver trasplantation (OLT) or with associated diseases contraindicating surgery; a single lesion (ⱕ5 cm) or two lesions ⬍3 cm each; Child-Pugh class A or B; absence of portal thrombosis or extrahepatic metastasis evaluated by computed tomography (CT) and chest x-ray; platelet count ⬎40,000/mm3; prothrombin time ratio ⬎40%. The mean follow-up was 23 ⫾ 15 months (range 1– 68 months). We used a commercial ultrasound (US) scanner with a 3.5-MHz real-time convex or annular probe and guide device. Sterile 95% ethanol was injected through either a 22-gauge spinal needle, 17.7 cm long, or a 22-gauge modified spinal needle, 20 cm long, with multiple holes. Alcohol was injected at a dose of 1–10 mm per session with one or more injections during the same session. The patients were

2922

Orlando et al.

treated every day or three times per week. We treated lesions ⬍2 cm in three to six sessions, and lesions of 2–5 cm in 10 to 15 sessions. The ethanol volume was calculated according to the formula of Shiina et al. (8). Before starting the procedure, all the patients were staged with dynamic CT scan, AFP, and chest x-ray. The mean time elapsed between HCC diagnosis and the starting of PEI was 1 month. Evaluation of Response and Follow-Up Protocol One month after the end of PEI cycle, a CT scan was repeated and the AFP level was measured. Tumor necrosis was considered when the lesion appeared as a hypoattenuated nonenhancing area on the CT scan (9), whereas tumor persistence was diagnosed when enhancing areas were demonstrated. When there was evidence of tumor persistence, a second cycle of PEI was indicated. If the tumor was still viable after a second cycle, treatment was deemed to have failed. The follow-up protocol included clinical assessment, measurements of liver function by Child-Pugh score (10), AFP level, US examination every 3 months, and abdominal CT scan every year. Local recurrence after total necrosis by CT scan was defined as any sign of progression in the treated tumor on a follow-up CT, such as the development of tumor staining at the margin of the tumor or enlargement of the tumor. When an enlargement of the size of lesion or a new lesion was observed during US or an increase of ⱖ100 ng/ml AFP levels with respect to previous assay was found, a CT scan was performed. If local recurrence was detected, PEI was repeated. A separate new lesion was considered to have developed when there was the appearance of another focal lesion distinct (i.e., without continuity) from the treated one. In this case US-guided biopsy was performed and if HCC was detected, a new cycle of PEI was undertaken. Patients who developed diffuse HCC were not retreated. Statistical Evaluation Continuous variables were expressed as mean ⫾ SD. The ␹2 test was performed as appropriate, with all p values being two-tailed. The primary outcome of the study was overall and tumorfree survival. Secondary outcomes were the development of separate new lesions and of local recurrence. The KaplanMeier method (11) was used to estimate the survival, the occurrence of separate new lesions, and local recurrence. Differences in the survival rate were assessed by the logrank test. The observed survival time was the interval between starting PEI and death or the final follow-up. For local recurrence and for the development of separate new lesions, the tumor-free time was the interval between starting PEI and the appearance of the tumor. In this analysis, the follow-up of patients dying without recurrence was censored at the time of death, and these patients were classified as disease-free. The following variables were considered for univariable

AJG – Vol. 95, No. 10, 2000

analysis: age, sex, bilirubin level, albumin level, prothrombin time ratio, alkaline phosphatase, baseline tumor size, number of lesions, basal platelets count, basal AFP level, and presence of esophageal varices at endoscopy. Variables with a p value at univariable analysis ⬍0.05 were included in the final multivariable model. Ascites assessed by ultrasound and encephalopathy were not assessed as prognostic factors because only a few patients had these complications at baseline (five patients with mild ascites and one with encephalopathy). The Cox proportional hazards model (12) was used to assess the overall survival, tumor-free survival, appearance of separate new lesions, and local recurrence in a multiple regression analysis. All analyses were conducted with the Statistical Analysis System, version 6.08 (SAS, Cary, NC). All p values were two-tailed, and all confidence intervals (Cis) were 95%. We verified that the most important issue of the Cox model (i.e., the assumption of proportional hazard) was not violated (13). On the basis of the final model, the cumulative estimated survival probability or so-called survivorship function S (t,z) corresponding to any combination of the variables z ⫽ z1. . .zp was computed: S (t,z) ⫽ A0(t)exp(␤⬘X) in a separate new test set of 105 consecutive patients who were treated by PEI in another center (14). This test set was a cohort of consecutive patients enrolled in the Vimercate Hospital Center (Milan) which was included in a previous published multicenter study (15). The estimated probability of survival were compared with a Kaplan-Meier plot of the observed data. The estimated (E) number of patients who died were computed from the survivorship function according to the following equation: ⫺log [A0(t)exp(␤⬘X)]. The difference between the estimated (E) and the observed (O) number of patients who died was tested by the one sample log-rank test according to the equation:

␹2 ⫽ (O1 ⫺ E1)2/E1 ⫹ (O2 ⫺ E2)2/E2 ⫹ . . . ⫹ (On ⫺ En)2/En.

RESULTS The clinical characteristics of the patients included in the study are shown in Table 1. A total of 49 deaths occurred: 22 patients in Child-Pugh A class, and 27 patients in ChildPugh B class. The causes of death were liver failure in 22 patients, GI bleeding in eight, hemoperitoneum in four, cachexia in two, unknown in three, and non–liver-related in 10 patients. Among the 49 deaths, 21 patients died without an active tumor disease (i.e., without separate new lesions or local recurrence). The overall survival and the survival according to Child-Pugh class are shown in Figure 1. The overall survival rates were 89% (CI 81–94), 63% (CI 52– 73), and 43% (CI 31–55) at 1, 2, and 3 yr, respectively. Survival was significantly longer among Child-Pugh A class patients (96% [CI 88 –99], 78% [CI 64 – 87], and 63% [CI 47–75] at 1, 2, and 3 yr, respectively) than among Child-

AJG – October, 2000

Ethanol Injection for Hepatocellular Carcinoma

Table 1. Baseline Features of 115 Patients Treated With Percutaneous Ethanol Injection Patient (n) Mean age (yr) Sex (M/F) Anti-HCV ⫹ (n) HbsAg ⫹ (n) ␣-Fetoprotein (ng/ml) Platelets ⱖ100,000/mm3 ⬍100,000/mm3 Serum albumin (g/L) Prothrombin activity (%) Serum bilirubin (mg/dl) Alkaline phosphatase (UI/L)* Presence of esophageal varices (n) Ascites Absent Mild No. of lesion Single lesion Two lesions Tumor size (cm) ⱕ3 ⬎3–5

Child-Pugh A

Child-Pugh B

81 64.8 ⫾ 7.4 61/20 75 6 129 ⫾ 552

34 62.5 ⫾ 5.5 21/13 31 3 238 ⫾ 766

33 44 36.5 ⫾ 3.9 84 ⫾ 13 1.0 ⫾ 0.5 194 ⫾ 150 32

9 21 30.2 ⫾ 4.9 65 ⫾ 12 2.0 ⫾ 1.5 212 ⫾ 158 15

81 0

29 5

66 15

30 4

48 33

20 14

Data are given as means ⫾ standard deviations. * Alkaline phosphatase normal value: 170 UI/L.

Pugh B class patients (73% [CI 54 – 85], 35% [CI 19 –52], and 12% [CI 3–27] at 1, 2, and 3 yr, respectively) (p ⬍ 0.0001 by the log-rank test).

2923

After univariable analysis, baseline albumin levels, platelet counts, and prothrombin activity had a p value ⬍0.05 and were included in the final model. Baseline albumin levels came out as the only significant independent predictors for overall survival (Table 2). The same prognostic factor was identified when tumor-free survival (HR 7.17, 95% CI 2.9 – 17.7) and non–tumor-free survival (HR 3.61, 95% CI 1.84 – 7.09) were separately assessed. A total of 40 patients developed new lesions. The 1-, 2-, and 3-yr rate of occurrence of separate new lesions was 21%, 30%, and 41% respectively (Fig. 2). We did not find variables significantly associated with the appearance of separate new lesion. In 94 patients (82%) a complete tumor necrosis, as evaluated by CT scan, was achieved in 73 patients after one cycle and in 21 patients after two cycles. In 20 patients local recurrence was observed. The 3-yr cumulative local recurrence rate was 23%. The basal tumor size (⬍3 cm and ⬎3 cm) was the only predictive factor that was significantly associated with local recurrence (HR 2.78, 95% CI 1.53– 5.04). Prospective Validation of the Survival Prognostic Model A validation of the survival prognostic tool was carried out with an external, independent set of 105 patients with HCC who underwent PEI in another center with the same inclusion criteria. The baseline features of this test set are shown in Table 3. The predicted individual cumulative probabilities of survival (survivorship functions) for the new group of

Figure 1. Overall survival curves of all the 115 patients treated with percutaneous ethanol injection (solid lines) for hepatocellular carcinoma and according to Child-Pugh class A (dotted lines) and Child-Pugh class B (broken lines).

2924

Orlando et al.

AJG – Vol. 95, No. 10, 2000

Table 2. Adjusted Relative Risk of Overall Survival in 115 Patients Treated With PEI for Hepatocellular Carcinoma Variable

Code

␤

SE

p Value

RR (95% CI)

Serum albumin levels

Group 1: ⬎35 g/L Group 2: 35–28 g/L

1.37

0.28

0.0001

3.95 (2.26–6.91)

Model ␹2 ⫽ 30.21 with 3 df, p ⬍ 0.0001; adjusted for baseline platelets count and prothrombin activity. CI ⫽ confidence interval; PEI ⫽ percutaneous ethanol injection RR ⫽ relative risk; SE ⫽ standard error.

patients were estimated. Patients were then divided into two groups according to their baseline albumin level: group 1, albumin levels ⬎35 g/L; group 2, albumin levels 35–28 g/L. The average predicted and observed probabilities of survival for the two groups were compared with Kaplan-Meier plots (Fig. 3). The difference between observed (O) and expected (E) number of patients who died were not significant in either of the albumin groups: group 1 (n ⫽ 58), observed 15, expected 9.6; group 2 (n ⫽ 47), observed 30, expected 23.6 (␹2⫽ 4.76, df ⫽ 2, p ⫽ 0.09).

DISCUSSION No randomized trials have been performed to evaluate the effectiveness of PEI in the treatment of small HCC. Although a randomized trial is difficult to carry out, cohort studies with analysis of prognostic factors may be useful to identify subgroups of patients in whom treatment is effective (16). The results of these studies are more reliable if confirmed in an external series, and need to be validated before these studies can be considered definitive. In our study, the 2- and 3-yr overall survival rates were 63% and 43%. These findings are quite similar to those of

the randomized trial of Ohnishi et al. (17) comparing PEI and percutaneous acetic acid injection, in which the survival rates of patients treated by PEI were 63% and ⬍40%, respectively. The 3-yr survival of Child-Pugh class A patients in our series (63%) was comparable with that of other series, ranging from 66% (5) to 87% (4). This rate is clearly higher than the 3-yr survival of untreated patients reported in two others studies (ranging from 21% to 26%) (1, 2). Only Barbara et al. (18) reported a lengthy survival (63%) for untreated Child-Pugh class A patients; however, their series was not consecutive and the number of patients was small. Despite the limits of an indirect comparison between treated and untreated patients, we believe that the available evidence is sufficient to conclude that PEI may prolong survival in small HCC associated with Child-Pugh class A cirrhosis. Among the three laboratory parameters of the Child-Pugh score (i.e., albumin, bilirubin, and prothrombin time), only the albumin level was found to be significant at multivariable analysis, suggesting that, in addition to cancer-related features (number and size of lesions), liver function of the

Figure 2. Rate of occurrence of separate new lesions in patients with hepatocellular carcinoma treated with percutaneous ethanol injection.

AJG – October, 2000

Ethanol Injection for Hepatocellular Carcinoma

Table 3. Clinical Characteristics of the External Independent Set of 105 Patients Patients (n) Mean age (y) Sex (M/F) ␣-Fetoprotein (ng/ml mean value) Serum albumin (g/L) Prothrombin activity (%) Serum bilirubin (mg/dl) Ascites Absent Mild No. of lesions Single lesion Two lesions Tumor size (cm) ⱕ3 ⬎3–5

Child-Pugh A

Child-Pugh B

87 64 ⫾ 7.6 75/12 150.6 ⫾ 431.8

18 63 ⫾ 6.5 10/8 262.7 ⫾ 857.1

38.1 ⫾ 4.3 75 ⫾ 14 1.06 ⫾ 0.5

32 ⫾ 4.2 60 ⫾ 14 2.19 ⫾ 0.8

87 0

12 6

61 26

15 3

23 64

4 14

Data are given as means ⫾ standard deviations.

underlying cirrhosis must be carefully evaluated before selecting patients for PEI. The poor 3-yr survival rate for Child-Pugh B class patients in our series (12%; CI 3–27%) is lower than that observed in other series, in which the 3-yr survival ranged from 41% (1) to 53% (4) and is similar to a previous untreated series in our center (2). It is difficult to explain the

2925

different results, considering that the rate of necrosis and of recurrence was comparable in all studies. However, in our series, Child-Pugh class B patients with normal albumin had a good rate of survival, indicating that this subgroup could benefit from PEI. In our study (and at variance with others), apart from serum albumin levels, no other variables such as size and number of lesions or basal AFP values emerge as independent predictors for better survival. The reliability of our prognostic model is strengthened by the validation performed in an external series of patients with the same features. Until now, only one other paper, using multivariable analysis, has analyzed predictors of survival of HCC treated by PEI (7). The authors identified baseline albumin levels, AFP values, and number of lesions as predictors of good clinical outcome. However, in this paper, the imbalance between the high number of variables included in the model (i.e., 13) and the small number of events observed (i.e., 28) could lead to a false-positive result. Many others studies (5, 19 –23) have analyzed variables associated with a better prognosis, and the following have been identified: Child-Pugh class, size and number of lesions, AFP ⬍200 ng/ml, and ascites. However, none of these studies take into account a confounding analysis, either by a multivariable model or by stratification. It is important to emphasize that, despite the satisfactory fit of our prognostic model to the external independent set of

Figure 3. Observed (solid lines) and estimated (broken lines) survival functions for the two groups of patients divided according to levels of baseline albumin. Group 1: albumin levels ⬎35 g/L; number of patients (N1 ⫽ 58), number observed dying (O1 ⫽ 15), number expected to die (E1 ⫽ 9.6). Group 2: albumin levels 35–28 g/L; number of patients (N2 ⫽ 47), number observed dying (O2 ⫽ 30), number expected to die (E2 ⫽ 23.6).

2926

Orlando et al.

data, further prospective studies are needed to confirm our results. The high rate of new lesions observed in our study and in other series (3, 7, 17) suggests that PEI cannot be considered a curative treatment of HCC associated with cirrhosis. It is important to underscore that we did not observe any difference in the development of separate, new lesions between the two Child-Pugh classes. This observation strengthens the relevance of the liver function as a predictor of survival. The high rate of recurrence is also observed in resected patients (3), indicating the palliative role of PEI and surgery. No data are available on the recurrence after transarterial embolization treatment. Moreover, this treatment has been used in nonsurgical HCC with large and/or multiple lesions, and there is no evidence that it prolongs survival (24, 25). Only in OLT has a low rate of recurrence (26, 27) been observed, suggesting a curative role of this technique. In fact, in an Italian series of HCC treated by OLT (27), the actuarial survival and the recurrence-free rate at 4 yr were 85% and 92%, respectively, for small HCC, and were 50% and 59% for HCC 3–5 cm, which is clearly better than the results obtained by PEI. Finally, the local recurrence was favored only by the initial size of the lesion. This result is in agreement with the observations of Vilana et al. and Ishii et al. (6, 28): tumors measuring ⬍3 cm seem to respond better to PEI. It is likely that a complete tumor necrosis can be achieved with acetic acid, as suggested by Ohnishi et al. (17), who found a significant lower rate of local recurrence by using acetic acid compared to PEI in the treatment of small HCC. In conclusion, our data suggest that patients with HCC and good liver function assessed by albumin level may benefit from PEI treatment despite basal AFP levels, initial tumor size (ⱕ5 cm), and number of lesions (ⱕ2). However, the high rate of new lesions suggests that PEI cannot be considered a curative treatment of HCC associated with cirrhosis, and underscores the relevance of OLT as the only curative treatment. However it is important to remember that the waiting list for OLT is often long and that OLT is not usually indicated in patients ⬎65 yr. Therefore, in these two groups of patients, PEI is indicated.

ACKNOWLEDGMENTS The authors thank Jeremy Stuart Randalls for the English revision of the manuscript, Maria Pia Bonini for technical assistance during PEI, and Lorenzo Tese`, Vincenzo Calistro, and Franco Valenza for CT performed. We are indebted to Professor Alberto Morabito (Istituto di Statistica Medica e Biometria, University of Milan) for advice. Reprint requests and correspondence: Ambrogio Orlando, Divisione di Medicina Interna, Ospedale V.Cervello, Via Trabucco 180, 90146 Palermo, Italy. Received Dec. 3, 1999; accepted May 8, 2000.

AJG – Vol. 95, No. 10, 2000

REFERENCES 1. Livraghi T, Bolondi L, Buscarini L, et al. No treatment, resection and ethanol injection in hepatocellular carcinoma: A retrospective analysis of survival in 391 patients with cirrhosis. J Hepatol 1995;22:522– 6. 2. Orlando A, Cottone M, Virdone R, et al. Treatment of small hepatocellular carcinoma associated with cirrhosis by percutaneous ethanol injection. A trial with a comparison group. Scand J Gastroenterol 1997;32:598 – 603. 3. Castells A, Bruix J, Bru C, et al. Treatment of small hepatocellular carcinoma in cirrhotic patients: A cohort study comparing surgical resection and percutaneous ethanol injection. Hepatology 1993;18:1121– 6. 4. Lencioni R, Bartolozzi C, Caramella D, et al. A. Treatment of small hepatocellular carcinoma with percutaneous ethanol injection. Analysis of prognostic factors in 105 western patients. Cancer 1995;76:1737– 46. 5. Livraghi T, Bolondi L, Lazzaroni S, et al. Percutaneous ethanol injection in the treatment of hepatocellular carcinoma: A study on 207 patients. Cancer 1992:926 –9. 6. Vilana R, Bruix J, Bru C, et al. Tumour size determines the efficacy of percutaneous ethanol injection for the treatment of small hepatocellular carcinoma. Hepatology 1992;16:352–7. 7. Castellano L, Calandra M, Del Vecchio Blanco C, et al. Predictive factors of survival and intrahepatic recurrence of hepatocellular carcinoma in cirrhosis after percutaneous ethanol injection: Analysis of 71 patients. J Hepatol 1997;27:862–70. 8. Shiina S, Tagawa K, Unuma T, et al. Percutaneous ethanol injection therapy for neoplasms located on the surface of the liver. AJR 1990;155:507–9. 9. Ebara M, Kita K, Sugiura N, et al. Therapeutic effect of percutaneous ethanol injection on small hepatocellular carcinoma: Evaluation with CT. Radiology1995;195:371–7. 10. Pugh RNH, Murray-Lyon IM, Dawson JL, et al. Transection of the esophagus for bleeding oesophageal varices. Br J Surg 1963;60:646 – 64. 11. Kaplan EI, Meier P. Non-parametric estimation from incomplete observation. J Am Stat Assoc 1958;53:457– 81. 12. Cox DR. Regression models and life tables (with discussion). J R Statist Soc B 1972;34:187–220. 13. Andersen PK. Testing goodness-of-fit methods for the proportional hazards regression model. Biometrics 1992;38:6777. 14. Breslow NE. Analysis of survival data under the proportional hazards model. Int Stat Rev 1975;1:45–58. 15. Livraghi T, Giorgio A, Marin G, et al. Hepatocellular carcinoma and cirrhosis in 746 patients: Long-term results of percutaneous ethanol injection. Radiology 1995;197:101– 8. 16. Rosati R, Lee K, Califf R, et al. Problems and advantages of an observational data base approach to evaluating the effect of therapy on outcome. Circulation 1982;65(suppl II):11–28. 17. Ohnishi K, Yoshioka H, Ito S, et al. Prospective randomized controlled trial comparing percutaneous acetic acid injection and percutaneous ethanol injection for small hepatocellular carcinoma. Hepatology 1998;27:67–72. 18. Barbara L, Benzi G, Gaiani S, et al. Natural history of small untreated hepatocellular carcinoma in cirrhosis: A multivariable analysis of prognostic factors of tumour growth rate and patients survival. Hepatology 1992;16:132–7. 19. Lencioni R, Pinto F, Armillotta N, et al. Long-term results of percutaneous ethanol injection therapy for hepatocellular carcinoma in cirrhosis: A European experience. Eur Radiol 1997; 7:514 –9. 20. Pompili M, Rapaccini GL, De Luca F, et al. Risk factors for intrahepatic recurrence of hepatocellular carcinoma in cirrhotic patients treated by percutaneous ethanol injection. Cancer 1997;79:1501– 8.

AJG – October, 2000

21. Ebara M, Ohto M, Sugiura N, et al. Percutaneous ethanol injection for the treatment of small hepatocellular carcinoma. Study of 95 patients. J Gastroenterol Hepatol 1990;5:616 –26. 22. Giorgio A, Tarantino L, Francica G, et al. Percutaneous ethanol injection under sonographic guidance of hepatocellular carcinoma in compensated and decompensated cirrhotics patients. J Ultrasound Med 1992;11:587–95. 23. Ishii H, Okada S, Nose H, et al. Predictive factors for recurrence after percutaneous ethanol injection for solitary hepatocellular carcinoma. Hepatogastroenterology 1996;43:938 – 43. 24. Bruix J, Llovet J, Castells A, et al. Transerterial embolization versus symptomatic treatment in patients with advanced hepatocellular carcinoma: Results of a randomized controlled trial in a single institution. Hepatology 1998;27:1578 – 83.

Ethanol Injection for Hepatocellular Carcinoma

2927

25. Groupe d’Etude et de Traitment du Carcinome Hepatocellulaire. A comparison of lipiodol chemoembolization and conservative treatment for unresectable hepatocellular carcinoma. N Engl J Med 1995;332:1256 – 61. 26. Llovet JM, Bruix J, Fuster J, et al. Liver transplantation for small hepatocellular carcinoma: The tumor-node-metastasis classification does not have prognostic power. Hepatology 1998;27:1572–7. 27. Mazzaferro V, Regalia E, Doci R, et al. Liver transplantation for the treatment of small hepatocellular carcinoma in patients with cirrhosis. N Engl J Med 1996;334:693–9. 28. Ishii H, Okada S, Nose H, et al. Local recurrence of hepatocellular carcinoma after percutaneous ethanol injections. Cancer 1996;77:1792–96