Prognostic factors after liver resection for hepatocellular carcinoma with hepatitis B virus-related cirrhosis: The surgeon's role in survival

Results The significant risk factors for decreasing both the overall and disease-free survival of patients were: (1) ascites volume of more than 500 ml; (2) prothrombin time of more than 4 s; (3) serum AFP of more than 400 ng/ml; (4) tumor distribution in two lobes; (5) vascular invasion; (6) capsule absence; and (7) blood transfusion of more than 600 ml. Moreover, female gender and operation time of more than 5 h are risk factors of tumor recurrence but not for the patients' overall survival. The 3-year survival rate decreased from 100% to 0 as the number of risk factors in the patients increased from zero to four or more. Patients who had two or more preoperative risk factors were poor candidates for liver resection, with a 3-year survival rate of 8.5%. Conclusions The survival of HCC patients with HBV-related cirrhosis after liver resection depends on preoperative liver reserve, tumor status and blood transfusion. Tumor status cannot be altered; however, the surgeon can do a great favor to the prognosis of patients by minimizing bleeding and blood transfusion. Patients with two or more preoperative risk factors should be cautiously selected for liver resection. Keywords Surgery Risk factor Hepatocellular carcinoma Hepatitis B virus Cirrhosis Surgeon Introduction Hepatocellular carcinoma (HCC) is the third most common cause of cancer-related death in the whole world, with over 500 000 new cases diagnosed yearly. 1 Liver cirrhosis is the most frequent risk factor for HCC, since about 85%–90% of the cases occur in cirrhotic livers. 2 The status of liver function significantly determines the clinical presentation and surgical management of HCC, as well as the outcome of the patients. Therefore, several papers have concluded different clinical presentations and risk factors affecting the outcome of patients in cirrhotic and noncirrhotic patients, respectively. Hepatitis B virus (HBV) and hepatitis C virus (HCV) are the two most important factors in hepatocarcinogenesis. Due to the difference in the pathogenesis of HBV- and HCV-related HCC, the clinical presentation and outcomes of patients in HBV- and HCV-related HCC were different. 3,4 However, little is known about the significant risk factors affecting the prognosis of HCC patients with HBV-related cirrhosis. The highest incidence of HCC has been seen in China. 5,6 From the 1990s, HCC ranked second in cancer-caused death in China. 7 In Chinese patients with HCC, 40%–90% show positive hepatitis B surface antigen and most have coexisted liver fibrosis and cirrhosis with limited hepatic reserve. 5,6 Therefore, in an HBV endemic area, it was significant and interesting to investigate the risk factors affecting the outcomes of HCC patients with HBV-related cirrhosis, which could help to guide treatment decision and improve patient survival. Patients and methods Patient population HBV-related cirrhosis was defined in this study as cirrhosis in the liver parenchyma with positive hepatitis B surface antigen but negative hepatitis C antibody. From 1 October 1996 to 1 October 2006, 535 consecutive patients with liver cirrhosis and newly diagnosed HCC underwent liver resection in the First Affiliated Hospital of Medical College, Xi'an Jiaotong University. Thirty eight cases with positive hepatitis C antibody, 12 with both positive hepatitis C antibody and hepatitis B surface antigen and 27 with other factors were excluded from the study. In the 458 cases with HBV-related cirrhosis, 46 (10.0%) patients were lost in follow-up. The remaining 412 with HBV-related cirrhosis who underwent liver resection for HCC were enrolled in this study. Detailed clinicopathologic and surgical data of the patients were retrospectively reviewed and are presented in Table 1 . Liver resection was the treatment of choice whenever the patients were operable on the basis of general condition, tumor characteristics and location, liver function tests, and volume of future remnant liver. Diagnosis and parameters definition All patients were diagnosed by preoperative computed tomography (CT) of the liver and some of them by intraoperative biopsy if CT was inconclusive and confirmed by pathological examination after surgery. Liver function was measured by the Child–Pugh classification system, and staging of tumor was evaluated by the American Joint Committee on Cancer (AJCC) tumor, nodes, metastasis (TNM) staging system. The results of biochemical tests at diagnosis were recorded for analysis. The sizes of the tumors are measured by preoperative CT and/or postoperative pathological examination. Ascites volume was calculated for the overall volume drained within 1 week before the surgery and during the operation. Operation time was defined as the duration between the beginning of skin incision and the accomplishment of incision suture. Anatomic resection was defined as the complete removal of at least one Couinaud segment containing the tumor, while nonanatomic resection was defined as removal of the tumor plus an edge of the nonneoplastic liver parenchyma. Perioperative blood transfusion volume included the volume of transfused whole blood and/or packed erythrocytes in the operation and within 1 week after surgery. The indication for and volume of blood transfusion were based on the blood pressure, hemoglobin and hematocrit levels of the patients. Follow-up After discharge, patients were followed up every 3 months by alpha-fetoprotein (AFP) detection and upper abdominal CT in the first 3 years and then every 6 months after that. If recurrence was highly suspected, hepatic angiography was performed for verification. Patients attending analysis Overall survival was measured in 412 patients, including operative deaths, hospital deaths, and noncured patients, from the date of surgery to the time of last follow-up (October 2007) or death. Disease-free survival was evaluated in the group of 351 patients with curative resection who were discharged from the hospital (13 patients with hospital death and 48 with noncurative surgery due to residual tumor tissue in the resection margin were excluded). Variable analyzed and statistic analysis All data were expressed as mean ± SD. The analyses of overall survival and disease-free survival were calculated by the Kaplan–Meier method, and the differences in survival between groups were compared using the log-rank test. To determine the factors that may be associated with the prognosis of the patients, 21 factors were tested in the patients group. Among the 21 variables, age (<60 years vs. ≧60 years), gender, Child classification (A vs. B), ascites (<500 ml vs. ≧500 ml), hepatitis B virus early antigen status (positive vs. negative), ALT (<54 U/l vs. ≧54 U/l), AST (<54 U/l vs. ≧54 U/l), TBIL (<17.1 μmmol/l vs. ≧17.1 μmmol/l), ALB (<35 g/l vs. ≧35 g/l) and PT (<4 s vs. ≧4 s) are considered “host-related factors”. AFP (<400 ng/ml vs. ≧400 ng/ml), tumor size (≦5 cm vs. 5 cm), number (solitary vs. multiple), distribution (one lobe vs. two lobes), vascular invasion (present vs. absent), capsule (present vs. absent), and TNM stage (I vs. II–III) are grouped into “tumor-related factors” and operation time (<5 h vs. ≧5 h), resection mode (nonanatomic vs. anatomic), surgical margin (<10 mm vs. ≧10 mm) and perioperative blood transfusion (<600 ml vs. ≧600 ml) are considered “treatment-related factors”. The results of the univariate analysis helped to select significant prognostic factors ( p < 0.10). The selected significant variables of the prognostic factors were then used in the subsequent multivariate analysis using Cox's proportional hazards model. Statistical analysis was carried out using SPSS 13.0. p < 0.05 was considered statistically significant. Results Patient characteristics The median follow-up period was 21 months (range 1–120 months). Tumor recurrence was verified in 292 (70.87%) patients. Among them, 70 were treated by repeat hepatic resection; 80 by transcatheter arterial chemoembolization (TACE) and chemotherapeutics infusion; 16 by percutaneous ethanol injection (PEI) and 38 by radiofrequency ablation (RA); and the other 26 patients by two or more treatments mentioned above. Still, 62 patients with recurrent HCC accepted no treatment due to severe general condition and poor liver function. The remnant liver was the most frequent site of recurrence (204/292, 69.86%). Other sites of recurrence included lung, bone, lymph node, adrenal gland, brain and pancreas. Prognostic factors for overall survival by univariate analysis Fourteen factors were selected as significant prognostic risk factors by the univariate analysis. The significant prognostic host-related factors were Child classification, ascites volume, serum AST level and albumin level, and prothrombin time (all p < 0.01). The significant prognostic tumor-related factors were serum AFP level, tumor size, tumor number and distribution, vascular invasion, capsule absence, and TNM stage (all p < 0.01). The significant prognostic treatment-related factors were operation time ( p = 0.01) and perioperative blood transfusion volume ( p < 0.01). Prognostic factors for overall survival by multivariate analysis Multivariate analysis was performed on 14 prognostic factors for overall survival that reached statistical significance in the univariate analysis ( p < 0.10). Ascites volume, PT, AFP, tumor distribution, capsule absence, vascular invasion and blood transfusion were selected as independent prognostic factors of poor overall survival, as shown in Table 2 . Prognostic factors for disease-free survival by univariate analysis Fifteen factors were selected as significant prognostic risk factors by the univariate analysis. The poor prognostic host-related factors were female gender ( p < 0.05), Child classification ( p < 0.01), ascites volume ( p < 0.01), and serum albumin level ( p < 0.01). The poor prognostic tumor-related factors were serum AFP level, tumor size, number and distribution, vascular invasion, capsule absence, and TNM stage (all p < 0.01). The poor prognostic treatment-related factors were operation time and perioperative blood transfusion volume (all p < 0.01). Prognostic factors for disease-free survival by multivariate analysis Multivariate analysis was performed on 15 prognostic factors for disease-free survival that reached statistical significance in the univariate analysis ( p < 0.10). As is shown in Table 3 , gender, ascites, PT, AFP, tumor distribution, absence of capsule, vascular invasion, operation time and blood transfusion were selected as independent prognostic factors of poor disease-free survival. Prognosis after hepatic resection Hospital death was 3.16% (13/412). The mean and median overall survival including hospital death after hepatic resection was 26 months and 22 months (range 1–120 months) respectively. The 1-, 3-, and 5-year overall survival rates after hepatic resection were 80%, 53% and 30%. The 1-, 3-, and 5-year disease-free survival rates after hepatic resection were 59%, 32% and 26%. We categorized the patients with different number of risk factors and compared the survival rate between them in Fig. 1 . The 3-year survival rate decreased from 100% to 0 as the number of risk factors in the patients increased from zero to four or more. Patients with two or more preoperative risk factors (ascites volume of more than 500 ml, PT of more than 4 s, serum AFP level of more than 400 ng/ml, tumor distribution in two lobes and vascular invasion) had a significantly lower 3-year survival rate than those with one or no preoperative risk factor (8.5% vs. 43% and 68%, p < 0.01, Kaplan–Meier analysis) ( Fig. 2 ). Discussion Although it is well known that different hepatitis virus backgrounds and cirrhosis could contribute to the outcome of HCC patients, little is known in judging the uncontrollable and controllable factors that affect the outcome of HCC patients with HBV-related cirrhosis undergoing liver resection. This study demonstrated that certain host-related factors (ascites and PT), tumor-related factors (tumor distribution, capsule formation, vascular invasion, serum AFP level) and blood transfusion were independent risk factors of poor overall survival and high recurrence rate of the patients. Also, gender and operation time were found to be significant factors affecting disease-free survival but not the overall survival of this group of patients. Therefore, the outcomes of the HCC patients with HBV-related cirrhosis largely depended on the preoperative hepatic functional reserve, tumor status and operation procedure. Because the tumor status cannot be altered, preoperative liver function can only be partially improved, surgeons can help improve prognosis if blood transfusion could be limited and operation time could be shortened by expertise. Tumor status and survival Stratifying the HCC patients by HBV status and cirrhosis that underwent liver resection, we found, in HBV-related cirrhotic patients, that tumor distribution in two lobes, capsule absence, vascular invasion and AFP level are significant factors affecting both the overall survival and disease-free survival. Similar to our findings, tumor status was mostly proved to be an independent risk factor for disease-free and overall survival of the patients 8–12 ; however, not in all cases. 13,14 It is reasonable to consider that biological characteristics and progression of the tumor in terms of tumor-related factors strongly signify the malignant behavior and metastatic capability of cancer, consequently affecting surgery results. Liver reserve and survival Since HCC often develops in the background of liver cirrhosis, hepatic function reserve is frequently suboptimal. Preoperative evaluation of liver function is, therefore, an important aspect of selecting patients with adequate function for liver resection. From 2006, the ICG clearance test, combined with other tests, was routinely taken for patients with HCC and limited liver function in our center. An ICG retention level of less than 40% at 15 min was generally used as a threshold for liver resection in our department; however, the volume of nontumorous liver and the TBIL level were co-considered. No hospital death occurred from 2006. Blood transfusion and survival Gantt et al. firstly reported the adverse effect of blood transfusion in cancer patients possibly by posttransfusion immunosupression. 15 However, some other authors failed to prove this effect in other tumors. 16–18 Blood transfusions in different cancers may have different significance contributing to recurrence and patient's survival. HCC is a rapidly growing, highly malignant and aggressive tumor with a poor prognosis compared with other cancers. Therefore, the contribution of blood transfusion to prognosis could be weakened and veiled to some extent. In our study, the blood transfusion volume was proved to be a significant factor determining the outcome of the patients, which was similar to other studies. 14 In contrast, Makino et al. found no correlation between perioperative blood transfusion and prognosis in gross patients but a significant enhancement of intrahepatic recurrence of HCC in those patients with portal vein invasion. 19 In general, blood transfusion has a nonspecific immunosuppression effect and consequently weakens the immune system of the host which plays an important role in limiting proliferation and dissemination of cancer cells. 19,20 Thus, the primary aim of a surgeon during hepatectomy should be the achievement of least bleeding and blood transfusion. The amount of blood transfusion is influenced by the host's conditions including liver functions, collateral veins, etc. and also by the expertise of the surgeon. Bleeding remains a major problem in liver resection. Bleeding may occur during mobilization of the tumor and transaction of the liver parenchyma from tearing of intrahepatic branches of the portal vein and hepatic vein, laceration of the tumor supplying arteries or iatrogenic tumor rupture. Bleeding in the operation always necessitates allogenic blood transfusion and may lead to the dissemination of cancer cells. Thus, rectifying the poor clotting system of patients preoperatively and controlling the loss of blood during an operation are significant for reducing blood transfusion. To reduce bleeding, firstly, adequate and wide exposure of the operative field is essential. Secondly, light and soft isolation of the tumor and adjacent tissue is vital in avoiding tumor rupture and vascular tearing. Thirdly, during tumor resection, intermittent use of the Pringle maneuver or selective hepatic vascular exclusion for 15 min with the following 5 min of unclamping is effective and safe for controlling bleeding in patients. Last but not the least, even with an ultrasonic dissector and argon beam coagulation, the obvious bleeding points should be still exactly sutured. Surgical margin and survival It was interesting to notice that a surgical margin less than 1 cm and nonanatomic resection did not significantly affect recurrence of the tumor and prognosis of the patients in our study ( p > 0.05). Although a resection margin of at least 1 cm was commonly obeyed by many surgeons, the role of the resection margin in contributing to the long-term survival of patients is still controversial. Some authors advocated a definite resection margin of over 1 cm and found that this could definitely prolong the overall survival of patients. 21–25 However, consistent with our study, other authors found no significant contribution of the surgical margin to tumor recurrence and the survival of patients. 12,13,26–28 Liver resection mode and survival Since HCC primarily spreads through intrahepatic portal venous branches or hepatic venous tributaries, it seems reasonable that anatomic resection eradicates potential venous tumor thrombi in the anatomic region. Therefore, some authors found in their study that anatomic resection achieved better disease-free and overall survival than limited resection. 12,29–31 However, consistent with some studies, 8,9,32 anatomic resection in our group did not show significant merits over nonanatomic resection in terms of the overall and disease-free survival of the patients ( p > 0.05). Recently, Yamashita et al. reported their 20 years of experience and demonstrated that anatomic resection should be recommended for noncirrhotic patients, and the limited resection group attained a better 5-year survival than the anatomic resection group, which proved the validity of limited resection for cirrhotic patients. 33 The largest retrospective study is from Eguchi et al. They summarized a nationwide Japanese database of 72 744 patients with HCC who underwent liver resection and found that anatomic resection attained better disease-free survival, especially when the size of the tumor ranges from 2 to 5 cm. 34 It was the difference of liver function of the patients enrolled in different studies that accounted for the disparity of the results. Many HCC patients have coexisted hepatitis and/or liver cirrhosis which greatly prevent major and anatomic resection. In addition, a cirrhotic liver is more prone to multicentric carcinogenesis. Even after an apparently curative resection, nearly 50% of the patients would still experience intrahepatic recurrence and the majority of these recurrences are multicentric locations and distant from the resection margin. 35–37 Thus, it implies that even anatomic resection and a wider resection margin may not prevent recurrence or prolong the overall survival of cirrhotic patients; conversely, it would increase the risk of postoperative liver failure. Risk-factor staging system and survival We identified seven independent risk factors affecting the survival of patients. The survival rate of the patients declined obviously as the number of risk factors increased. Patients bearing two or more of the five preoperative risk factors (ascites volume of more than 500 ml, PT of more than 4 s, serum AFP level of more than 400 ng/ml, tumor distribution in two lobes and vascular invasion) were shown to have a poor prognosis even after liver resection. Therefore, some salvage treatment should be considered for these patients. Moreover, a close postoperative recheck should be carried out and adjuvant therapy, i.e. TACE, PEI and RA, for these patients should be recommended. Conclusions Surgical technique is as important as preoperative liver reserve function and tumor status in terms of survival of the patients. Minimizing bleeding and blood transfusion help to improve the outcome of the patients. Patients with two or more preoperative risk factors were poor candidates for liver resection. Other adjuvant treatments could be considered. Even after surgery, a close postoperative recheck was more essential for these patients. Conflict of interest There is no conflict of interest. Acknowledgement We thank Dr. Yuan Shen in the Medical Statistical Department of Medical College, Xi'an Jiaotong University, for assistance with the statistical analysis. References 1 D.M. Parkin F. Bray J. Ferlay P. Pisani Estimating the world cancer burden: Globocan 2000 Int J Cancer 94 2 2001 153 156 2 H. Lang G.C. Sotiropoulos E.I. Brokalaki Survival and recurrence rates after resection for hepatocellular carcinoma in noncirrhotic livers J Am Coll Surg 205 1 2007 27 36 3 Q. Li H. Li Y. Qin W. Peter X. Hao Comparison of surgical outcomes for small hepatocellular carcinoma in patients with hepatitis B versus hepatitis C: a Chinese experience J Gastroenterol Hepatol 22 2007 1936 1941 4 Y.H. Huang J.C. Wu C.H. Chen Comparison of recurrence after hepatic resection in patients with hepatitis B vs. hepatitis C-related small hepatocellular carcinoma in hepatitis B virus endemic area Liver Int 25 2005 236 241 5 J. Ferlay F. Bray P. Pisani Globocan 2000: “cancer incidence, mortality and prevalence worldwide” 2001 IARC Press Lyon, France p. 3 6 X.S. Hao P.P. Wang K.X. Chen Twenty-year trends of primary liver cancer incidence rates in an urban Chinese population Eur J Cancer Prev 12 2003 273 279 7 S.W. Zhang L.D. Li F.Z. Lu Mortality of primary liver cancer in China from 1990 through 1992 Chin J Oncol 21 1999 245 249 in Chinese with English abstract 8 H.J. Park K.C. Koh M.S.C. Choi Analysis of risk factors associated with early multinodular recurrences after hepatic resection for hepatocellular carcinoma Am J Surg 192 2006 29 33 9 T. Ochiai T. Sonoyama D. Ichikawa Poor prognostic factors of hepatectomy in patients with resectable small hepatocellular carcinoma and cirrhosis J Cancer Res Clin Oncol 130 2004 197 202 10 B.H. Lang R.T. Poon S.T. Fan J. Wang Perioperative and long-term outcome of major hepatic resection for small solitary hepatocellular carcinoma in patients with cirrhosis Arch Surg 138 2003 1207 1213 11 T. Sonoyama T. Ochiai T. Hironaka H. Yamagishi Predictors of postoperative diffuse intrahepatic recurrence of hepatocellular carcinoma Hepatogastroenterology 50 2003 1078 1084 12 H. Imamura Y. Matsuyama E. Tanaka Risk factors contributing to early and late phase intrahepatic recurrence of hepatocellular carcinoma after hepatectomy J Hepatol 38 2003 200 207 13 H.G. Sim L.L. Ooi Results of resections for hepatocellular carcinoma in a new hepatobiliary unit ANZ J Surg 73 2003 8 13 14 S.T. Fan O.L. Ng R.T. Poon H. Yamagishi Hepatectomy for hepatocellular carcinoma: the surgeon's role in long-term survival Arch Surg 134 1999 1124 1130 15 C.L. Gantt Red blood cells for cancer patients Lancet 2 1981 363 16 J.H.W. Donohue S. Cha Perioperative blood transfusions do not affect disease recurrence of patients undergoing curative resection of colorectal carcinoma: a Mayo Clinic/North Central Cancer Treatment Group study J Clin Oncol 13 1995 1671 17 The Danish Breast Cancer Cooperative Group J.H. Eikhoff J. Anderson C. Laybourn Perioperative blood transfusion does not promote recurrence and death after mastectomy for breast cancer Br J Surg 78 1991 1358 1361 18 S.R. Velagapudi M. Frydenberg J.E. Oestering Homologous blood transfusion in patients with prostate cancer: no effect on tumor progression or survival Urology 43 1994 821 827 19 Y. Makino A. Yamanoi T. Kimoto On. El-Assal H. Kohno N. Nagasue The influence of perioperative blood transfusion on intrahepatic recurrence after curative resection of hepatocellular carcinoma Am J Gastroenterol 95 2000 1294 1300 20 H.J. Nielsen Detrimental effects of perioperative blood transfusion Br J Surg 82 1995 582 587 21 M. Shi R.P. Guo X.J. Lin Partial hepatectomy with wide versus narrow resection margin for solitary hepatocellular carcinoma: a prospective randomized trial Ann Surg 245 1 2007 36 43 22 M. Lise S. Bacchetti P. Da Pian D. Nitti L. Pilatip P. Pigato Prognostic factors affecting long term outcome after liver resection for hepatocellular carcinoma: results in a series of 100 Italian patients Cancer 82 1998 1028 1036 23 Liver Cancer Study Group of Japan Predictive factors for long term prognosis after partial hepatectomy for patients with hepatocellular carcinoma in Japan Cancer 74 1994 2772 2780 24 I. Ikai S. Arii M. Kojiro Reevaluation of prognostic factors for survival after liver resection in patients with hepatocellular carcinoma in a Japanese nationwide survey Cancer 101 2004 796 802 25 I.O. Ng E.C. Lai S.T. Fan M.M. Ng M.K. So Prognostic significance of pathologic features of hepatocellular carcinoma: a multivariate analysis of 278 patients Cancer 76 1995 2443 2448 26 R.T.P. Poon S.T. Fan O.L.I. Ng J. Wong Significance of resection margin in hepatectomy for hepatocellular carcinoma: a critical reappraisal Ann Surg 231 2000 544 551 27 K. Ouchi S. Matsubara K. Fukuhara T. Tominagat S. Matsuno Recurrence of hepatocellular carcinoma in the liver remnant after hepatic resection Am J Surg 166 1993 270 273 28 R. Izumi K. Shimizu T. Li Prognostic factors of hepatocellular carcinoma in patients undergoing hepatic resection Gastroenterology 106 1994 720 727 29 J.M. Regimbeau R. Kianmanesh O. Farges F. Dondero A. Sauvanet J. Belghiti Extent of liver resection influences the outcome in patients with cirrhosis and small hepatocellular carcinoma Surgery 131 3 2002 311 317 30 K. Hasegawa N. Kokudo H. Imamura Prognostic impact of anatomic resection for hepatocellular carcinoma Ann Surg 242 2005 252 259 31 M. Yamamoto K. Takasaki T. Ohtsubo H. Katsuragawa C. Fukuda S. Katagiri Effectiveness of systematized hepatectomy with Glisson's pedicle transection at the hepatic hilus for small nodular hepatocellular carcinoma: retrospective analysis Surgery 130 2001 443 448 32 K. Tanaka H. Shimada C. Matsumoto Anatomic versus limited nonanatomic resection for solitary hepatocellular carcinoma Surgery 143 2008 607 615 33 Y. Yamashita A. Taketomi S. Itoh Long-term favorable results of limited hepatic resections for patients with hepatocellular carcinoma: 20 years of experience J Am Coll Surg 205 1 2007 19 26 34 S. Eguchi T. Kanematsu S. Arii Comparison of the outcomes between an anatomical subsegmentectomy and a non-anatomical minor hepatectomy for single hepatocellular carcinomas based on a Japanese nationwide survey Surgery 143 2008 469 475 35 J. Belghiti Y. Panis O. Farges J.P. Benhamou F. Fekete Intrahepatic recurrence after resection of hepatocellular carcinoma complicating cirrhosis Ann Surg 214 1991 114 117 36 E. Adachi T. Maeda T. Matsumata Risk factors for intrahepatic recurrence in human small hepatocellular carcinoma Gastroenterology 108 1995 768 775 37 T. Kumada S. Nakano I. Takeda Patterns of recurrence after initial treatment in patients with small hepatocellular carcinoma Hepatology 25 1995 87 92