799 Paraneoplastic Syndromes in Patients with Hepatocellular Carcinoma in Taiwan Jiing-Chyuan Luo, M.D.1 Shinn-Jang Hwang, M.D.1 Jaw-Ching Wu, M.D., Ph.D.1 Chung-Pin Li, M.D.1 Linag-Tsai Hsiao, M.D.2 Chiung-Ru Lai, M.D.3 Jen-Huei Chiang, M.D.4 Wing-Yiu Lui, M.D.5 Full-Young Chang, M.D.1 Shou-Dong Lee, M.D.2 1 Division of Gastroenterology, Veterans General Hospital-Taipei and National Yang-Ming University School of Medicine, Taipei, Taiwan. 2 Department of Medicine, Veterans General Hospital-Taipei and National Yang-Ming University School of Medicine, Taipei, Taiwan. 3 Department of Pathology, Veterans General Hospital-Taipei and National Yang-Ming University School of Medicine, Taipei, Taiwan. 4 Department of Radiology, Veterans General Hospital-Taipei and National Yang-Ming University School of Medicine, Taipei, Taiwan. 5 Department of Surgery, Veterans General Hospital-Taipei and National Yang-Ming University School of Medicine, Taipei, Taiwan. BACKGROUND. Hepatocellular carcinoma (HCC) is the most common malignancy in Taiwan. Some patients may manifest paraneoplastic syndromes during the clinical course of the disease. In this study, the authors evaluated the clinical significance of these paraneoplastic syndromes, compared the prevalence of these syndromes between cases of hepatitis B virus (HBV)–related and hepatitis C virus (HCV)–related HCC, and estimated significant predictors associated with the syndromes. METHODS. Clinical data on 1197 HCC patients, including age, gender, Child–Pugh score, survival time, laboratory data (including liver biochemistry, hepatitis markers, and serum a-fetoprotein [AFP]), and tumor features (including tumor size, portal vein thrombosis, and histologic pictures), were retrospectively reviewed. RESULTS. A total of 232 of 1197 patients (19.4%) had paraneoplastic syndromes. HCC patients with paraneoplastic syndromes had significantly higher serum AFP; higher rates of initial main portal vein thrombosis, metastasis, and bilobal tumor involvement; larger tumor volume; and shorter survival than those without these syndromes. Patients with HBV-related HCC had a significantly higher prevalence of paraneoplastic syndromes than patients with HCV-related HCC (20.1% vs. 11.2%, P 5 0.005). In a stepwise multivariate logistic regression analysis, AFP .50,000 ng/mL and tumor volume .30% were significant predictive variables associated with the presence of paraneoplastic syndromes in HCC patients. CONCLUSIONS. HCC patients with paraneoplastic syndromes usually had higher levels of serum AFP and larger tumor volumes than those without. Patients with HBV-related HCC had a significantly higher prevalence of paraneoplastic syndromes than those with HCV-related HCC. Cancer 1999;86:799 – 804. © 1999 American Cancer Society. KEYWORDS: a-fetoprotein, erythrocytosis, hepatitis B virus, hepatitis C virus, hepatocellular carcinoma, hypercalcemia, hypercholesterolemia, hypoglycemia, paraneoplastic syndromes. H Supported by a grant from Veterans General Hospital-Taipei, Taiwan. Address for reprints: Shinn-Jang Hwang, M.D., Division of Gastroenterology, Department of Medicine, Veterans General Hospital-Taipei, 201 ShihPai Road, Section 2, Taipei, 11217, Taiwan. Received September 29, 1998; revision received March 8, 1999; accepted April 1, 1999. © 1999 American Cancer Society epatocellular carcinoma (HCC) is the most common malignancy in Taiwan. During its clinical course, patients may manifest a variety of paraneoplastic syndromes, including hypercholesterolemia, hypoglycemia, hypercalcemia, and erythrocytosis.1– 8 According to our previous reports, the incidences of these syndromes were 11.4% if hypercholesterolemia, 2.8% if hypoglycemia, 1.8% if hypercalcemia, and 2.5% if erythrocytosis.9 –12 Hepatitis B virus (HBV) infection is endemic in Taiwan, with a prevalence rate of 15–20%. In contrast, the prevalence of hepatitis C virus (HCV) infection is only 1–2% in the general Taiwanese population, although very high prevalence has been found in several townships.13,14 According to previous reports, hepatitis B surface antigen (HBsAg) and anti-HCV were positive in 74 –77% and 12–19% of HCC 800 CANCER September 1, 1999 / Volume 86 / Number 5 patients, respectively, in Taiwan.15,16 Here, we evaluate the clinical significance of HCC patients with paraneoplastic syndromes, compare the prevalence of these syndromes between HBsAg positive and antiHCV positive HCC patients, and assess whether differences in laboratory data and tumor features exist between the two groups. MATERIALS AND METHODS The clinical, laboratory, and imaging data were reviewed for 1472 consecutive new or follow-up patients diagnosed with HCC at the Division of Gastroenterology, Veterans General Hospital-Taipei, from January, 1991 to December, 1994. Of these patients, 1253 were enrolled in this study who met the following criteria: 1) histologic proof of HCC or 2) at least two typical HCC image findings (ultrasonography, computerized tomography, celiac angiography, or magnetic resonance imaging) along with a serum a-fetoprotein (AFP) level . 20 ng/mL. Finally, data from 1197 patients were analyzed after excluding patients with alcoholassociated cirrhosis and HCC as well as patients who lacked complete examination and data for analysis. We defined hypercholesterolemia as a serum cholesterol level . 250 mg/dL (two standard deviations [S.D.] above the mean value of age- and gendermatched, healthy controls) after excluding past history of diabetes mellitus, hypothyroidism, past history or family history of hypercholesterolemia, and evidence of extrahepatic biliary obstruction.9 We defined hypoglycemia as a plasma glucose , 60 mg/dL.17,18 Hypercalcemia was defined as a serum calcium level . 11.0 mg/dL after adjustment by serum albumin concentrations and excluding hyperparathyroidism and medication- and bony metastasis-induced hypercalcemia.11 Erythrocytosis was defined as a hemoglobin level . 16.7 gm/dL or a hematocrit level . 50% (two S.D. above the mean value of age- and gendermatched, healthy controls) after excluding secondary erythrocytosis due to chronic pulmonary disease, residence at high altitudes, and polycythemia vera.10 Underlying cirrhosis was diagnosed histologically or by characteristic image findings with the presence of ascites or esophageal varices. Patients with cirrhosis were given a score from 5 to 15 according to the Child–Pugh classification.19 Tumor volume was calculated from computerized tomographic films and was expressed as the percentage of tumor volume divided by total liver volume. The grade of differentiation and arrangement of tumor cells were assessed by a liver pathologist according to the classification of Edmondson and Steiner.20 The pathologist was not given any clinical information pertaining to the biopsy specimens. All clinical data, including age, gender, Child– Pugh scores, liver biochemistry (measured with a Hitachi model 736 automatic analyzer, Tokyo, Japan), prothrombin time, and complete blood count, were recorded when HCC was first diagnosed and at the time of development of the paraneoplastic syndromes. Serum AFP was measured by using a commercial kit (ELSA2-AFP; CIS Bio-International, Cedex, France) at each hospitalization after HCC was diagnosed. Initial serum AFP was defined as the AFP value when HCC was first diagnosed, and peak serum AFP was defined as the greatest AFP value measured during the clinical course of HCC. Anti-HCV was measured by using a second-generation enzyme immunoassay kit (Abbott Laboratories, Chicago, IL). The serum markers of HBsAg, antibody to HBV core antigen (anti-HBc), antiHBs, and antibody to hepatitis D virus (Abbott Laboratories) were recorded. Initial main portal vein (MPV) thrombosis was defined as evidence of MPV tumor invasion when HCC was first diagnosed. Distant metastases to extrahepatic regional lymph nodes or other organs were evaluated by imaging studies. Initial metastasis was defined as evidence of tumor spread upon diagnosis of HCC. Methods of therapy for HCC, including surgical resection of tumor, transcatheter arterial chemoembolization (TAE), sonoguided percutaneous ethanol injection, systemic chemotherapy, and survival time, were recorded. Data in the text and tables are expressed as the mean 6 S.D., the median, and/or the range. Serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and AFP were analyzed by using raw data and data after logarithmic transformation. Results were compared between groups by using the chi-square test or Fisher exact test, the Student t-test, or the Mann–Whitney test, depending on the type of data analyzed. Survival adjusted for therapy was analyzed by using the Kaplan–Meier method and was compared by using the log-rank method. Univariate, multivariate logistic regression using SPSS software (SPSS Inc., Chicago, IL) was performed to evaluate the predictive values of the patients’ clinical, laboratory, and tumor features associated with paraneoplastic syndromes. For all tests, only results with P , 0.05 were considered to be statistically significant. RESULTS Among the 1197 HCC patients for final analysis, the mean age was 62 6 12 years, with a range of 13– 89 years. Males were predominant, with a male-to-female-ratio of 6.8:1.0. Seven hundred sixty-three patients (63.7%) were HBsAg positive, 187 patients (15.6%) were anti-HCV positive, 55 patients (4.6%) were positive for both HBsAg and anti-HCV, 22 pa- Paraneoplastic Syndromes in HCC/Luo et al. 801 TABLE 1 Comparison of Clinical and Laboratory Data and Tumor Features between Hepatocellular Carcinoma Patients with and without Paraneoplastic Syndromes Characteristic HCC patients with paraneoplastic syndromes (n 5 232) HCC patients without paraneoplastic syndromes (n 5 965) Age (yrs) Gender (male:female) HBV:HCV related Mean Child–Pugh score Mean initial a-fetoprotein (ng/mL) Median (range) Mean peak a-fetoprotein (ng/mL) Median (range) Initial MPV tumor thrombosis (1:2) Initial metastasis (1:2) Tumor volume (%) Bilobal tumor involvement (1:2) Therapy for HCC (1:2) Tumor cell arrangement (trabecular:mixed:acinar) Tumor cell differentiation (Grade 1:2:3:4) Median survival (days) 61 6 14 211:21 154:21 6.8 6 2.1 122,084 6 280,188 2895 (3–1,621,700) 171,803 6 352,790 8701 (3–2,055,300) 60:172 61:171 47.0 6 25.1 120:112 100:132 36:11:3 12:33:7:2 152 63 6 11 833:132 609:166 6.8 6 2.2 23,719 6 106,116 196 (3–1,892,500) 35,695 6 167,960 459 (3–2,975,570) 125:840 112:853 20.1 6 16.2 353:612 523:442 131:42:17 30:121:46:7 634 P value 0.042 0.074 0.006 0.894 ,0.001 ,0.001 ,0.001 ,0.001 ,0.001 ,0.001 0.003 0.792 0.331 ,0.001 HCC: hepatocellular carcinoma; HBV: hepatitis B virus; HCV: hepatitis C virus; MPV: main portal vein; 1:2: positive:negative. Data are expressed as the mean 6 standard deviation. tients (1.9%) were positive for HBsAg and antibody to hepatitis D virus, 170 patients (14.2%) were negative for HBsAg and anti-HCV, and 121 of these 170 patients (71%) had serum anti-HBc. A total of 232 of 1197 patients (19.4%) had paraneoplastic syndromes during the clinical course of HCC, of whom 177 had a single paraneoplastic manifestation, and 55 had multiple paraneoplastic manifestations. The prevalence of hypercholesterolemia was 12.1%, hypoglycemia was 5.3%, hypercalcemia was 4.1%, and erythrocytosis was 3.1%. When comparing the clinical and laboratory data and the tumor features between HCC patients with and without paraneoplastic syndromes, HCC patients with paraneoplastic syndromes were significantly younger in age. They had higher rates of HBV infection than HCV infection, greater initial and peak serum AFP levels, higher rates of initial MPV thrombosis, higher rates of metastases and bilobar tumor involvement, and larger tumor volumes. They were less likely to receive or be suitable for HCC therapy, and they had shorter survivals than those without paraneoplastic syndromes (Table 1). There were no significant differences in gender distribution; mean Child–Pugh score (Table 1); serum levels of ALT, AST, and total bilirubin; or rate of cirrhosis between patients with and without paraneoplastic syndromes (data not shown). HCC patients with these syndromes tended to have greater tumor volumes. The mean percentage of tumor volume was 47% 6 25% with a range of 12–95%. One hundred and fifty-four of 763 patients (20.1%) with HBV-related HCC had paraneoplastic syndromes, including 97 (12.7%) with hypercholesterolemia, 46 (6.0%) with hypoglycemia, 32 (4.2%) with hypercalcemia, and 24 (3.1%) with erythrocytosis. Twenty-one of 187 patients (11.2%) with HCV-related HCC had paraneoplastic syndromes, including 14 (7.5%) with hypercholesterolemia, 3 (1.6%) with hypoglycemia, 3 (1.6%) with hypercalcemia, and 4 (2.1%) with erythrocytosis. Patients with HBV-related HCC had a significantly higher prevalence of hypercholesterolemia, hypoglycemia, and total paraneoplastic syndromes than those with HCV-related HCC (P 5 0.037, 0.006, and 0.005 respectively). When comparing the clinical and laboratory data and the tumor features between patients with HBVand HCV-related HCC with paraneoplastic syndromes, those with HBV-related HCC with paraneoplastic syndromes were significantly younger and had higher serum AFP levels than those with HCV-related HCC when the syndromes developed (Table 2). There were no significant differences in gender distribution, mean Child–Pugh score, rate of MPV thrombosis, metastases and bilobal tumor involvement, tumor volume, or survival between the two groups (Table 2). The serum levels of ALT, AST, and total bilirubin and the rate of cirrhosis were not significantly different between the two groups (data not shown). Two hundred fifty-eight of 1197 patients (22%) had adequate HCC tissue for histologic analysis, which included 54 patients with and 204 patients without paraneoplastic syndromes. Tumor cell arrangement 802 CANCER September 1, 1999 / Volume 86 / Number 5 TABLE 2 Comparison of Clinical and Laboratory Data and Tumor Features between Hepatitis B Virus- and Hepatitis C Virus-Related Hepatocellular Carcinoma Patients with Paraneoplastic Syndromes Characteristic HBV-related HCC with paraneoplastic syndromes (n 5 154) HCV-related HCC with paraneoplastic syndromes (n 5 21) Age (yrs) Gender (male:female) Mean Child–Pugh score Mean a-fetoprotein (ng/mL) Median (range) MPV tumor thrombosis (1:2) Metastasis (1:2) Tumor volume (%) Bilobal tumor involvement (1:2) Therapy for HCC (1:2) Tumor cell arrangement (trabecular:mixed:acinar) Tumor cell differentiation (Grade 1:2:3:4) Median survival (days) 58 6 14 137:17 7.8 6 2.4 182,218 6 350,400 3598 (3–1,621,700) 54:100 39:115 49.3 6 23.7 79:75 40:114 26:9:2 8:25:5:2 153 67 6 9 19:2 7.2 6 2.2 24,146 6 54,989 597 (4–165,330) 4:17 6:15 42.5 6 29.4 9:12 9:12 5:1:0 1:4:1:0 152 P value ,0.001 1.000 0.316 ,0.001 0.224 0.958 0.367 0.469 0.175 0.749 0.937 0.480 HBV: hepatitis B virus; HCC: hepatocellular carcinoma; HCV: hepatitis C virus; MPV: main portal vein. Data are expressed as the mean 6 standard deviation. showed a trabecular pattern in 167 patients, a mixed trabecular and acinar pattern in 53 patients, an acinar pattern in 20 patients, a compact pattern in 4 patients, and a scirrhous pattern in 14 patients. Clear cell variants were noted in 8 patients with paraneoplastic syndrome and in 27 patients without paraneoplastic syndrome. Tumor cell differentiation revealed Grade 1 in 42 patients, Grade 2 in 154 patients, Grade 3 in 53 patients, and Grade 4 in 9 patients. The tumor cell arrangement, the presence of clear cell variants, and tumor cell differentiation tended to show no significant difference between patients with and without paraneoplastic syndromes (Table 1) as well as between patients with HBV- and HCV-related HCC with paraneoplastic syndromes (Table 2). Age, gender, HBsAg positive status, anti-HCV positive status, serum AFP level, MPV thrombosis, metastases, bilobar tumor involvement, and tumor volume all were selected as independent variables in a univariate logistic regression analysis, with the presence of paraneoplastic syndrome as the dependent variable. The continuous variables were transformed to categoric variables, with the cut-off points determined by the receiver operating characteristic curve. In this analysis, AFP . 50,000 ng/mL, tumor volume . 30% of total liver volume, bilobar tumor involvement, initial MPV tumor thrombosis, and metastases were correlated significantly with the presence of paraneoplastic syndromes (Table 3). In a step-wise, multivariate logistic regression analysis, AFP . 50,000 ng/mL (odds ratio, 2.82; 95%; confidence interval, 1.82– 4.36; P , 0.001) and tumor volume . 30% (odds ratio, 3.23; 95%; confidence interval, 2.22– 4.68; P , 0.001) were significant predictive variables associated with the TABLE 3 Significant Predictive Variables Associated with Paraneoplastic Syndromes in Patients with Hepatocellular Carcinoma, Used in Univariate Logistic Analysis Characteristic Odds ratio 95% confidence interval P value a-Fetoprotein . 50,000 ng/mLa Tumor volume .30%a Bilobal tumor involvement Initial MPV tumor thrombosis Initial metastasis 4.9000 4.7533 1.8479 2.3299 2.2254 3.4208–7.0189 3.3883–6.6682 1.3842–2.4670 1.6438–3.3025 1.5999–3.0956 ,0.000 ,0.001 ,0.001 ,0.001 ,0.001 MPV: main portal vein. a Significantly predictive variables in multivariate analysis (see text for details). presence of these paraneoplastic syndromes in HCC patients. DISCUSSION HCC is the most common malignancy in Taiwan. In this study, the prevalence of HBsAg and anti-HCV positivity was 63.7% and 15.6%, respectively, in our HCC patients. However, the prevalence of HBV-related HCC in Taiwan began to decrease after universal HBV vaccination in neonates and children was implemented in 1984.21 Over the next 10 –20 years, HCVrelated HCC may be as prevalent in Taiwan as in Western countries. The prevalence of paraneoplastic syndromes in this study, including hypercholesterolemia, hypoglycemia, hypercalcemia, and erythrocytosis, was similar to what was found in our previous reports.9 –12 Different rates of incidence of these syndromes were re- Paraneoplastic Syndromes in HCC/Luo et al. ported in various studies and areas, and this variation in incidence may be due to the use of different definitions and values for paraneoplastic syndromes. The mechanisms by which these syndromes develop have been studied previously. Except in cases of hypercholesterolemia, hormone-like substances have been found in these patients. Hypoglycemia has been related to the overproduction of a substance with insulin-like activity, insulin growth factor II (IGF-II).17,22 It has been suggested that the cause of hypercalcemia is related to the overproduction of parathyroid-related protein (PTHrP), which interacts with PTH receptors.8,23 Elevation of serum erythropoietic activity has been seen in some HCC patients with erythrocytosis.10,24 The loss of feedback control and an increase in the active form of 3-hydroxy-3-methylglutaryl coenzyme A reductase (the major rate-limiting enzyme in cholesterol biosynthesis in mammalian cells) were postulated in HCC-related hypercholesterolemia.25,26 Defective receptors for chylomicron remnants on the surface of HCC cells, resulting in autonomous cholesterol biosynthesis, also has been reported.27 According to our results, HCC patients with paraneoplastic syndromes had significantly higher serum AFP levels and greater tumor volumes than those without these syndromes. We also found that patients with HBV-related HCC had a significantly greater prevalence of these syndromes than those with HCVrelated HCC. The greater tumor volume was similar in patients with both HBV- and HCV-related HCC with the syndromes, but those with HBV-related HCC with the syndromes had significantly higher serum AFP levels than patients with HCV-related HCC with the syndromes. In a step-wise, multivariate logistic regression analysis, we found that large tumor volume and a high serum AFP value were independent factors in predicting the appearance of paraneoplastic syndromes. In addition, previous reports have shown that HCC patients with paraneoplastic manifestations were related to the release of various functional hormonelike substances (PTH-rP, IGF-II, and erythropoietin).8,10,17,22–24 The levels of both serum AFP and hormone-like substances have decreased in parallel after effective therapy for HCC, including surgery and TAE, but they rose again with tumor recurrence or progression.8,10 According to these findings, we believe that a large tumor volume is necessary for the manifestation of paraneoplastic syndromes in patients with HCC.9 –11,17 We also suggest a hypothesis that high serum AFP levels in patients with HCC with paraneoplastic syndromes indicate enhanced biological activity of the tumor cells for the production of hormone-like substances, thus inducing paraneoplastic manifestations.9,17 803 Overexpression of the AFP gene in HCC cells contributes to the elevation in serum AFP levels.28,29 The correlation between AFP level and grade of tumor cell differentiation has been reported incoherently. Some reports showed that the presence of AFP tended to correlate with poorly differentiated HCC or high grade HCC,28,30,31 whereas other reports revealed no correlation between AFP level and tumor grade.32,33 Conversely, two reports stated that patients with both Grade 1 and Grade 4 HCC were associated with lower serum AFP levels.34,35 The correlation between cellular differentiation of HCC and the presence of paraneoplastic syndromes has not been studied well to date. According to the current study and our previous reports,9,10 there was no significant correlation between differentiation of HCC and paraneoplastic manifestations. However, further studies with more biopsy cases are needed to confirm this finding. In this study, it was observed that HCC patients with paraneoplastic syndromes had shorter survival, and they were less likely to be candidates for effective therapy, because they usually had larger tumor burdens and higher rates of MPV thrombosis and metastases. Early diagnosis and effective therapy for patients with less advanced HCC may prevent tumor growth and progression. However, the question of whether early and effective therapy can decrease the rate of paraneoplastic manifestations deserves further long term, prospective studies. In conclusion, HCC patients with paraneoplastic syndromes tended to have higher serum AFP levels and greater tumor volumes than those without these syndromes. Moreover, patients with HBV-related HCC had a significantly higher prevalence of these syndromes than patients with HCV-related HCC. REFERENCES 1. 2. 3. 4. 5. 6. 7. Okuda K, Nakashima T. Primary carcinomas of the liver. In: Berk JE, editor. Gastroenterology, 4th edition. Philadelphia: W. B. Saunders, 1985:3315–76. McFadzean AJS, Yeung RTT. Further observations on hypoglycemia in hepatocellular carcinoma. Am J Med 1969;47: 220 –35. Brownstein MH, Ballard HS. Hepatoma associated with erythrocytosis. Am J Med 1966;40;204 –10. Kan YW, McFadzean AJS, Todd D, Tso SC. Further observations on polycythemia in hepatocellular carcinoma. Blood 1961;18:592– 8. Goldberg RB, Bersohn I, Kew MC. Hypercholesterolemia in primary cancer of the liver. S Afr Med J 1975;23:1464 – 6. Oldenburg WA, van Heerden JA, Sizemore GW, Abbound CF, Sheedy PF. Hypercalcemia and primary hepatic tumors. Arch Surg 1982;117:1363– 6. Yeung RTT, Yeung DCY. Hypoglycemia associated with lipid accumulation in primary hepatocellular carcinoma. Cancer 1973;32:1482–9. 804 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. CANCER September 1, 1999 / Volume 86 / Number 5 Yen T-C, Hwang S-J, Wang C-C, Lee S-D, Yeh S-H. Hypercalcemia and parathyroid hormone-related protein in hepatocellular carcinoma. Liver 1993;13:311–5. Hwang S-J, Lee S-D, Chang C-F, Wu J-C, Tsay S-H, Lui W-Y, et al. Hypercholesterolemia in patients with hepatocellular carcinoma. J Gastroenterol Hepatol 1992;7:491– 6. Hwang S-J, Lee S-D, Wu J-C, Chang C-F, Lu C-L, Tsay S-H. et al. Clinical evaluation of erythrocytosis in patients with hepatocellular carcinoma. China Med J (Taipei) 1994;53:262–9. Hwang S-J, Chan C-Y, Wu J-C, Wang Y-J, Lee S-D, Tsai Y-T. et al. Clinical study of hypercalcemia in patients with hepatocellular carcinoma. Chinese J Gastroenterol 1989;6:182– 6. Hwang S-J, Wu J-C, Lee S-D, Tsay S-H, Lin H-C, Hwang Y-S, et al. Clinical and pathological study of hepatocellular carcinoma patients with hypoglycemia. Chinese J Gastroenterol 1988;5:38 –9. Chen D-S, Kuo GC, Sung J-L, Lai M-Y, Sheu J-C, Chen P-J. et al. Hepatitis C virus infection in an area hyperendemic for hepatitis B and chronic liver disease: the Taiwan experience. J Infect Dis 1990;162:817–22. Lu S-N, Chue P-Y, Chen I-L, Wang J-H, Huang J-F, Peng C-F. et al. Incidence of hepatitis C infection in a hepatitis C endemic township in southern Taiwan. Kaohsiung J Med Sci 1997;13:605– 8. Lee S-D, Lee F-Y, Wu J-C, Hwang S-J, Wang S-S, Lo K-J. The prevalence of anti-hepatitis C virus among Chinese patients with hepatocellular carcinoma. Cancer 1992;69:342–5. Chuang W-L, Chang W-Y, Lu S-N, Su W-P, Lin Z-Y, Chen S-C, et al. The role of hepatitis B and C viruses in hepatocellular carcinoma in a hepatitis B endemic area. A casecontrol study. Cancer 1992;69:2052– 4. Wu J-C, Daughaday WH, Lee S-D, Hsiao TS, Chou C-K, Lin H-D, et al. Radioimmunoassay of serum IGF-I and IGF-II in patients with chronic liver diseases and hepatocellular carcinoma with or without hypoglycemia. J Lab Clin Med 1988; 112:589 –94. Service FJ. Hypoglycemic disorders. N Engl J Med 1995;332: 1144 –52. Pugh RNH, Murry-Lyon M, Dawson JL, Pietroni MC, Williams R. Transection of esophagus for bleeding esophageal varices. Br J Surg 1973;60:646 –9. Edmondson HA, Steiner PE. Primary carcinoma of the liver, a study of 100 cases among 48900 necropsies. Cancer 1954; 7:462–503. Chang M-H, Chen C-J, Lai M-S, Hsu H-M, Wu T-C, Kong M-S et al. Universal hepatitis B vaccination in Taiwan and the incidence of hepatocellular carcinoma in children. N Engl J Med 1997;336:1855–9. Frizelle FA, Pfiefer MV. Paraneoplastic hypoglycemia treated with somatostatin analogue SMS 201-995. Eur J Surg Oncol 1996;22:546 –50. 23. Panesar NS, Au KM, Leung NWY, Shek CC, Swaminathan R. Nephrogenous cyclic AMP in primary hepatocellular carcinoma patients with or without hypercalcemia. Clin Endocrinol 1991;35:527–32. 24. Kew MC, Fisher JW. Serum erythropoietin concentrations in patients with hepatocellular carcinoma. Cancer 1986;58: 2485– 8. 25. Siperstein MD, Gyde AM, Morris HP. Loss of feedback control of hydroxymethylglutaryl coenzyme A reductase in hepatomas. Proc Natl Acad Sci U S A 1971;68:315–7. 26. Kawata S, Takaishi K, Nagase T, Ito N, Matsuda Y, Tamura S, et al. Increase in the active form of 3-hydroxy-3-methylglutary coenzyme A reductase in human hepatocellular carcinoma: possible mechanism for alteration of cholesterol biosynthesis. Cancer Res 1990;50:3270 –3. 27. Danilewitz MD, Herrera GA, Kew MC, Mendelsohn D, Barnes S, Alexander CB, et al. Autonomous cholesterol biosynthesis in murine hepatoma. Cancer 1984;54:1562– 8. 28. Peng S-Y, Lai P-L, Chu J-S, Lee P-H, Tsung P-T, Chen D-S, et al. Expression and hypomethylation of a-fetoprotein gene in unicentric and multicentric human hepatocellular carcinoma. Hepatology 1993;17:35– 41. 29. Ohguchi S, Nakatsukasa H, Higashi T, Ashida K, Nouso K, Ishizaki M, et al. Expression of alpha-fetoprotein and albumin genes in human hepatocellular carcinoma: limitations in the application of the genes for targeting human hepatocellular carcinoma in gene therapy. Hepatology 1998;27: 599 – 607. 30. Alpert E, Hershberg R, Schur PH, Isselbacher KJ. a-fetoprotein in human hepatoma: improved detection in serum, and quantitative studies using a new sensitive technique. Gastroenterology 1971;61:137– 43. 31. Vogel CL, Primack A, McIntire KR, Carbone PP, Anthony PP. Serum alpha-fetoprotein in 184 Ugandan patients with hepatocellular carcinoma. Cancer 1974:33:959 – 64. 32. Kashala LO, Kalengayi MMR, Essex M. Alpha-fetoprotein synthesis in human hepatocellular carcinoma: correlation with hepatitis B surface antigen expression. Cancer Invest 1992;10:513–22. 33. Izumi R, Shimizu K, Kiriyama M, Hashimoto T, Urade M, Yagi M, et al. Alpha-fetoprotein production by hepatocellular carcinoma is prognostic of poor patient survival. J Surg Oncol 1992;49:151–5. 34. The Liver Cancer Study Group of Japan. Primary liver cancers in Japan. Cancer 1980;45:2663–9. 35. Matsumoto Y, Suzuki T, Asada I, Ozawa K, Tobe T, Honjo I. Clinical classification of hepatoma in Japan according to serial changes in serum alpha-fetoprotein levels. Cancer 1982;49:354 – 60.