1195 Flow Cytometric DNA Analysis of Cirrhotic Liver Cells in Patients with Hepatocellular Carcinoma Can Provide a New Prognostic Factor Savino Rua, M.D.' Alberto Comino, M . D . ~ Adriana Fruttero, Ph.0.' Pierfederico Torchio, M.D? Hedayat Bouzari, M . D . ~ Stefan0 Taraglio, M . D . ~ Bruno Torchio, M.D? Lorenzo Capussotti, M.D." ' Department of Pathological Anatomy and His-. tology, City Hospital, Cuneo, Italy. * Department of Pathological Anatomy and His-. tology, City Hospital, Mondovi, Italy. Institute of Statistics, University of Milan, Mi.. lam, Italy. Department of Surgery, Hospital, Mauriziano "Umberto I st.", Torino, Italy. Department of Pathological Anatomy and Histology, Hospital, Mauriziano "Umberto I st.", Torino, Italy. BACKGROUND. DNA flow cytometry of hepatocellular carcinoma (HCC) cells has been investigated in many studies, but, to the best of our knowledge, there are no data on DNA analysis of cirrhotic parenchyma around the HCC. In this study, cell kinetics and ploidy of parenchymal cells around HCC were performed to ascertain if this would predict the possibility of recurrence in the cirrhotic areas. METHODS. The DNA content of 93 cases of HCC and of cirrhotic liver around the tumor nodules was analyzed by flow cytometry. Ploidy and proliferative index of HCC and cirrhotic liver were compared with macroscopic, histologic, and clinical features of each case and linked with the behavior of these tumors. Survival curves were assessed according to the Kaplan-Meier method. A multivariate analysis based on Cox proportional hazards regression model was performed on cases of diploid cirrhosis cells in which the S-phase fraction was evaluable. RESULTS. The univariate analysis of survival suggested significant roles for age, number of intrahepatic nodules, Edmondson-Steiner's classification, portal invasion, vascular invasion, presence of necrosis, hepatitis B surface antigen, u-fetoprotein, Child's score, ploidy, and S-phase fraction of HCC cells. The DNA analysis of the cirrhotic cells showed that polyploidy was dramatically reduced in patients with HCC, compared with normal hepatocytes, and aneuploid clones were present among diploid cells. High S-phase fraction of cirrhotic cells and Child-Pugh classification were the strongest independent parameters affecting the tumor behavior in this study. CONCLUSIONS. The results of this study suggest that S-phase fraction of cirrhotic liver parenchyma may be employed as a new parameter in the prognostic evaluation of HCC patients. Cancer 1996; 78:1195-1202.0 1996Arnericun Cancer Society. KEYWORDS: DNA flow cytometry, S-phase fraction, hepatocellularcarcinoma, cinhosis. M Addiress for reprints: Savino Rua, M.D., Department of Pathological Anatomy and Histology, City Hospital, Via Coppino 26, 12100 Cuneo, Italy. Received November 13, 1995; revisions received February 20, 1996 and May 31, 1996; accepted May 31, 1996. 0 'I996 American Cancer Society ost hepatocellular carcinomas (HCC)arise in cirrhotic patients. Hepatocyte proliferation is significantlycorrelated to the development of HCC in these patients.' In fact, those with a high cell proliferation run an increased risk of developing HCC and may require a differentiated follow-up.2Moreover, precancerous lesions, such as hyperplastic or adenomatous nodules, which may be present in cirrhotic livers, could give rise to tumor recurrence in surgically treated HCC ~atients.3.~ The prognosis of patients with HCC depends both on morphologic factors, such as tumor size and differentiation,presence of satellite nodules, and vascular invasion, and on clinical factors, such as patient's age, general clinical condition, liver function, Child-Pugh cirrhosis class, and/ or kind of surgical treatment?'* Although there have been many studies carried out on prognostic 1196 CANCER September 15,1996 / Volume 78 / Number 6 TABLE 1 Clinical Features of 93 Cirrhotic Patients with HCC Variable Sex Male Female Alcohol abuse Present Absent HbsAg Positive Negative Child-pugh class A B C Esophageal varices Present Absent Pattern of the cirrhosis Macronodular Microimacronodular Micronodular Diffuse fibrosis TABLE 2 Comparison of Different Pathologic Features of HCC and DNA Content No. Variable a2 11 88.2% 11.8% 59 34 63.4% 36.6% 23 70 24.7% 75.3% 59 30 4 63.4% 32.2% 4.4% 42 51 45.2% 54.9% 1 1.1% 79.6% 6.4% 12.9 74 6 12 HCC:hepatocellular carcinoma: HbsAg: hepatitis B surface antigen. No. of nodules Single Multiple Maximal tumor size” 5 3 cm >3 cm Distance from the cut level 4 0 mm 210 mm Macroscopy Expanding Spreading Diffuse Mixed Microscopy Trabecular Compact Polymorphous Endmondson classification I I1 111 N significance of DNA flow cytometry of HCC,9,’3to the best of our knowledge there are none on the DNA analysis of cirrhotic parenchyma around the HCC. Only Anti et al.I4 looked at flow cytometry in needle biopsies in a small group of patients with hepatic lesions (some with HCC),although they did not correlate their findings with survival. Therefore, our first objective in this study was to evaluate whether DNA flow cytometric analysis and the S-phase fraction of the cirrhotic liver around HCC nodules may predict survival in surgically treated patients. Portal invasion Present Absent Vascular invasion Present Absent Necrosis Present Absent u-feroprotein SMedian >Median No. Diploid Ploidy aneuploid Tetraploid 34 59 11 33.3% 22 66.7% 17 35.4% 31 64.6% 6 6 50. 50. 29 64 14 42.4% 19 57.6% 8 16.7% 40 83.3% 7 5 58. 41. 47 46 16 48.5% 17 51.5% 25 52.1% 23 47.9% 6 6 50. 50. 61 19 2 7 23 4 1 1 32 10 1 2 6 5 0 1 50. 41. 91 1 1 31 96.9% 22 52 18 1 8 16 8 1 22 71 79.3% 13.7% 3.5% 3.5% 71.2% 22.2% 2.2% 4.4% 8. 48 98.0% 1 2.0% 0 - 12 100 0 0 12 30 6 0 25.0% 62.5% 12.5% - 2 6 4 0 16. 50. 33. 8 24.2% 25 75.7% 10 79.2% 38 20.8% 4 8 33. 66. 42 51 19 57.6% 14 42.4% 27 56.2% 21 43.8% 5 7 41. 58. 38 55 16 48.5% 17 51.5% 29 60.4% 19 39.6% 10 8. 2 16. 47 46 19 57.6% 14 42.2% 21 43.7% 2 i 56.2% 7 5 0 - 1 3.1% 24.2% 48.5% 24.2% 3.1% 58. 41. ‘2‘: 10.78 P < 0.01. MATERIALS AND METHODS Patients Ninety-three cirrhotic patients were studied. All patients had undergone curative hepatectomy for HCC between August 1983 and July 1993 at the “Umberto I st.” Hospital Turin, Italy. Curative hepatectomy was defined as no visible tumor and/or tumor demonstrable during an intraoperative examination in the residual liver. The surgical procedures applied were anatomical resections in 77 patients (82.8%)and nonanatomical resections in 16 (17.2%).Two tumors (2.1%) had been treated with chemoembolization before surgery. In five patients the histograms were unsuitable for analysis and were, therefore, excluded from this study. Histologic examination of the tumor and liver parenchyma with cirrhosis was performed on surgical specimens embedded in paraffin after fixation in formalin for all patients. All cases were evaluated macroand microscopically according to the main criteria described by Japanese a~thors’~-’’ and graded according to Edmondson and Steiner.” The following variables were analyzed: age of patient at the time of diagnosis, alcohol abuse, hepatitis B surface antigen (HBsAg),Child-Pugh’s class, esophageal varices, cirrhosis pattern, number of nodules, maximal tumor size, tumor distance from the margin of resection, macroscopic pattern, Edmondson-Stein- DNA Content and Prognosis in HCC1Rua et al. TABLE 3 Ploidy of Cirrhotic Parenchymal Cells around HCC Dip1oid Aneuploid Tetraploid 87.7% 11.2% 1.1% 81 11 1 HCC: liepaiacellular carcinoma TABLE 4 Comparison between S-Phase Fraction and Other Prognostic Variables in 82 Cases of Diploid Cirrhotic Liver S-phase fraction Variable Modality No. (%) >4% No. (%) Age 4 2 >62 4 B t C 15 (36) 28 (70) 33 (63) 10 (331 27 (64) 12 (30) 19 (37) 20 1671 54% Child x2 = 9.70 P < 0.01 x2 = 6.90 P < 0.01 er’s grading, portal and/or vascular invasion, necrosis, a-fetoprotein (AFP),nuclear DNA content of both tumor cells and of cirrhotic liver cells, and survival. HCV testing performed after 1990 was performed on 18 patients. Flow Cytornetry Flow cytometry was performed on tumor and cirrhotic liver speciniens taken from the same paraffin block, including at least 5 to 10% of leukocytes andlor fibrocytes which were used as internal diploid reference cells. DNA content analysis was carried out on paraffin embedded specimens, from which 100 p thick sections were cut immediately adjacent to those used for histology.” Sections were dewaxed, rehydrated and, after treatment in 0.5% pepsin solution and washing, stained with propidium iodide according to a previously described method.20-zzTo avoid cell clumping, measurement samples were first checked at light microscopy. The measurements were performed using a FACScan flow cytometer (Becton Dickinson, San Jose, CAJ with a 488-nm argon ion laser. Histograms were divided into two groups, diploid, i.e., one GOlGl peak with a second which represented the G2lM-phase relative to the first, and nondiploid, i.e., those presenting different peaks, according to Joensuu and Klemi.23 A DNA index (DI) was calculated by dividing the modal channel number of the peak with higher DNA content by that of the peak with lower DNA content. 1197 Peaks determining a DI value between 1.85 and 2.15 were considered to be in the G2/M region. Cases with a peak in the G2/M region and a number of cells greater than 15%24were considered to be tetraploid. In diploid cases the S-phase fraction, the G2/Mphase fraction, and the proliferative index (sum of S and G2lM fractions) were calculated from DNA histograms by DNA cell-cycle analysis software (Becton Dickinson). Data Analysis The following parameters were considered for analytical purposes: sex, age, ploidy (hepatoma and cirrhosis), S-phase fraction (hepatoma and cirrhosis), maximum tumor size, number of intrahepatic nodules, portal invasion, vascular invasion, tumor distance from the margin of resection, HBsAg, AFP, Child-Pugh’s classification. Survival time was the dependent variable. Survival curves were assessed according to the Kaplan-Meier methodz5using the PROC LIFETE5T of SAS.26The curves were plotted at monthly intervals. Differences among curves were assessed according to the log-rank test.27 To control the effect of potential confounders, a multivariate analysis based on Cox’s proportional hazards regression model2*was performed on cases of diploid cirrhosis where the S-phase fraction was evaluable. The variables analyzed included age ( 5 62 and > 62 yrs), sex, S-phase fraction ( 5 1%,> 4%), tumor distance from the margin of resection (> 10 mm, < = 10 mm), number of intrahepatic nodules (single, multiple), portal invasion, vascular invasion, ChildPugh’s classification (2 levels: A and B + C), AFP ( 5 25, > 25 ng/mL), and necrosis. The exponentiation of the coefficient estimated from the regression model can be expressed as relative risk (RR) of dying during the follow-up period of subjects in the exposed category of each variable, compared with the reference category, after allowing for the other factors entered in the model. The PROC PHREG of SAS was used for Cox r e g r e s ~ i o nNinety-five .~~ percent confidence intervals (95%CI) of RRs were calculated by exponentiating the estimated coefficient plus or minus 1.96 times its standard error. RESULTS The principal clinical characteristics of the patients were studied (Table 1). All patients were Italian, 11 women (11.8%)and 82 men (88.2%),age 26 to 78 years (median: 62 yrs). 1198 CANCER September 15,1996 / Volume 78 / Number 6 TABLE 5 Kaplan-Meier's Estimates of the Cumulative Proportion of Survival (and Corresponding Standard Errors) according to the Considered Prognostic Factors Mos from surgery Variable Modality No. 12 36 96 F Sex Male Female 562 yrs 72 11 39 44 26 57 31 52 0.71 (0.05) 1 0.62 (0.08) 0.86 (0.05) 0.88 (0.05) 0.68 (0.06) 0.90 (0.05) 0.65 (0.07) 0.42 (0.06) 0.51 (0.16) 0.23 (0.07) 0.62 (0.08) 0.52 (0.10) 0.36 (0.07) 0.68 (0.09) 0.29 (0.06) 0.27 (0.05) 0.51 10.16) 0.16 (0.07) 0.47 (0.12) 0.42 (0.10) 0.24 (0.11) 0.51 (0.16) 0.21 10.06) 0.20 21 46 16 65 18 51 32 0.90 (0.06) 0.74 (0.06) 0.56 (0.12) 0.77 (0.05) 0.67 (0.11) 0.82 (0.05) 0.63 (0.09) 0.56 (0.11) 0.44 (0.08) 0.25 (0.11) 0.49 (0.07) 0.17 (0.09) 0.57 (0.07) 0.22 (0.08) 0.45 10.14) 0.32 (0.08)b 43 40 47 36 61 22 23 41 58 25 39 44 18 21 72 11 39 33 0.77 (0.06) 0.73 (0.07) 0.85 (0.05) 0.61 (0.08) 0.77 (0.05) 0.63 (0.10) 0.79 (0.06) 0.71 (0.07) 0.86 (0.05) 0.48 (0.10) 0.85 (0.07) 0.66 (0.06) 0.88 (0.08) 0.81 (0.09) 0.75 (0.05) 0.73 (0.13) 0.90 (0.05) 0.58 (0.09) 0.56 (0.08) 0.33 (0.08) 0.60 (0.08) 0.22 (0.07) 0.52 (0.07) 0.18 (0.09) 0.60 (0.08) 0.26 (0.08) 0.57 (0.07) 0.12 (0.07) 0.49 (0.10) 0.39 (0.07) 0.59 (0.13) 0.44 (0.12) 0.44 (0.06) 0.40 (0.16) 0.54 (0.08) 0.33 (0.08) Age >62 yrs Tumor size 5 3 cm >3 cm Number of nodules Edmonson-Steiner classification Portal invasion Necrosis Distance from the margin of resection Vascular invasion HBsAg a-fetoprotein Single Multiple I I1 111-IV Absent Present Absent Present 210 mm <10 mm Absent Present Negative Positive 525 ng/mL >25 n g h L Child's score A Hepatoma's ploidy B-C Diploids t Tetra Aneuploids Hepatoma's S-phase' 55% >5% Cirrhosis ploidy Diploids t Tetra Aneuploids Cirrhosis' S-phase' 54% >4% 0.01 0.24 0.01 0.02 0.01 0.35 (0.10) 0.17 (0.09) 0.45 (O.ll)h 0.01 0.47 (O.llih 0.01 0.43 (0.12) 0.19 10.07) 0.40 (0.10) 0.06 (0.06) 0.42 (0.12) 0.23 (0.08) 0.41 (O.lO)b 0.01 0.39 (0.10) 0.28 (0.08) 0.53 (0.13)b 0.19 0.32 (0.08)h 0.88 0.47 (0.08) 0.13 (0.10) 0.01 0.01 0.01 0.01 0.16 Sunjival analysis performed on 83 of93 patients (10 died in the postoperative period. "Log rank test. " N o patient reached 96 months of follos-up. ' Only diploids. HBsAg: heoatitis B surface antieen. Flow Cytometry The examined HCC were 35.5% diploid tumors, 46.2% aneuploid, and 12.9% tetraploid; 5.4% HCC showed a mosaic DNA distribution of aneuploidy. In the statistical analysis, 12 cases of HCC, which showed a DNA index of 2 (tetraploid tumor) were included in the diploid group. The DNA content of HCC was compared with conventional prognostic variables (Table 2). Tumor size was significantly related to aneuploidy. In fact, 40 of 64 lesions larger than 3 cm were aneuploid (62.5%), whereas only 8 of 29 tumors measuring < 3 cm were aneuploid (27.5%) ( P < 0.01). There was no significant relationship between ploidy, patient's age, sex, HBsAg, AFP, Child-Pugh class, number of intrahepatic nodules, portal invasion, vascular invasion, or necrosis. Moreover, there was not any apparent correlation between ploidy and HCV testing in the group of 18 patients evaluated. The majority of analyzed cases of peritumoral cir- DNA Content and Prognosis in HCC/Rua et al. 1 1199 S-phase fraction of cirrhotic cells and patient's age, sex, HBsAg, AFP, tumor size, Kanai and Edmondson classification, number of intrahepatic nodules, portal invasion, vascular invasion, and necrosis. I -; 0 12 24 36 48 60 72 84 96 108 120 Months h r n surgery FIGURE 1. Kaplan-Meier's estimates of the cumulative curve of survival after hepatectomy of all the 83 patients considered in the univariate analysis. 0 12 24 36 48 60 72 84 96 108 I 120 Months h o r n surgery FIGURE 2. Cumulative proportion of survival after hepatectomy in relation to S-phase fraction of cirrhotic parenchymal cells. A significant difference in survival between the curves was found ( P < 0.01 by the log rank test). rho tic tissue were diploid (87.7%),11 (11%)were aneuploid, and just 1 was tetraploid (Table 3). The diploid cases were selected to compare their S-phase fraction with two other important prognostic factors such as age and Child-Pugh classification. When S-phase fraction was compared with age, 64% of the patients younger than 62 years (64%) showed high S-phase value (>4%), whereas only 30% of the older patients showed the same values ( P < 0.01). Child-Pugh classes were significantly related to S-phase fraction of cirrhotic liver, in fact, 36.5% of patients with Child-Pugh Class A and 66.7% of patients with Child-Pugh Class B or C showed a high S-phase fraction ( P < 0.01) (Table 4). There was no significant relationship between Survival analysis Analysis was made on 83 cases because 10 patients (10.7%)who died in the postoperative period (within 2 mos) were excluded from the analysis of survival. Follow-up was continued until December 1994 or until a patient's death and ranged from 2 to 120 months (median, 30). Forty-nine patients died of neoplastic disease. The univariate analysis of survival showed that sex, tumor size, hepatoma S-phase fraction, hepatoma, and cirrhosis ploidy were not significantly correlated with survival (Table 5). The cumulative proportion of survival was 0.31 with 5 patients still alive 96 months after diagnosis (Fig. 1).Cumulative proportion of survival in relation to S-phase fraction of diploid cirrhotic liver was significantly higher in low (54%) versus high (> 4%) S-phase fraction cases ( P < 0.01 by the log-rank test) (Fig. 2). Cox Proportional Hazard Model Multivariate analysis demonstrated that the most predictive variables in decreasing order were Child-Pugh score, S-phase fraction of cirrhotic liver, and presence of necrosis. Conversely, significant prognostic indicators by univariate analysis, such as age, number of intrahepatic nodules, portal invasion, vascular invasion, HbsAg and AFP were unsignificant by multivariate analysis (Table 6). DISCUSSION In this study we demonstrated that the S-phase fraction of cirrhotic liver with diploid DNA content might predict the survival of patients with HCC. In our series the cirrhotic liver around HCC nodules was diploid in 91.1% of the cases, aneuploid in 7.8%, and tetraploid in 1.1%.The DNA analysis of the cirrhotic parenchyma showed that polyploidization was dramatically reduced in patients with HCC and aneuploid clones were present among diploid cells. It is known that normal human liver contains polyploid hepatocytes, and that the majority of them have two diploid nuclei. Polyploidization is thought to reflect increasing degrees of irreversible hepatocellular differentiati~n.~' In contrast, significantly elevated fractions of diploid hepatocytes and a reduction in the polyploid populations have been reported in humans and experimentally induced hepatocellular carcinoma^.^' Moreover in liver affected by chronic hepatitis and cirrhosis the diploid/ 1200 CANCER September 15,1996 / Volume 78 / Number 6 TABLE 6 Association between Prognostic Factors and Probability of Death Variable Modality RR Sex Male Female 562 yrs >62 yrs Negative Positive A B-C Single Multiple 1.00 1.50 1.00 1.22 1.00 2.39 1.oo 3.08 1.00 0.77 210 mm <10 mm 1.00 1.82 1.00 2.94 1.00 1.93 1.00 1.59 1.00 2.00 1.oo 3.61 Age HBsAg Child Number of nodules Distance from the margin of resection Necrosis cu-fetoprotein Vascular invasion Portal invasion S-phase of cirrhosis Absent Present 525 n g h L >25 ng/mL Absent Present Absent Present 54% >4% CI Chi-square - 0.47-4.8 0.47 - 0.46-3.2 0.16 - 0.98-5.80 3.69 - 1.44-6.62 8.32 0.23-2.52 0.19 - 0.79-4.15 1.98 - 1.32-6.57 6.94 - 0.95-3.93 3.28 - 0.61-3.93 1.60 - 0.68-5.82 0.90 - 1.51-8.63 8.28 RR: relative risk; CI: 95% confidence interval; HBsAg: hepatitis B surface antigen. The degree of association is expressed as RR and corresponding CI estimated fitting Cox multivariate proportional hazard model that includes as covariates gender and factors significantly associated with survival by univariate analysis (Table 51, polyploid ratios increased, whereas the hepatocyte binuclearity decreased this trend was even more marked in euploid nodules (premalignant and malignant), in which the S-phase fraction was significantly greater than that of the normal 1 i ~ e r . l ~ On the grounds of the hypothesis of a relationship between the development of carcinoma and an increased cell proliferation, presumably by an increased rate of random mutations,’ we analyzed the follow-up of patients with HCC associated to diploid cirrhotic liver in relation to high and low S-phase. In effect, the survival of patients with HCC which developed in a cirrhotic liver with a diploid DNA content and a high S-phase fraction was significantly reduced. In agreement to our findings, Tarao et al.’ demonstrated that HCC developed in a 3-year period in 64.3% of the cirrhotic patients with high-DNA synthesis activity, and only in 14.3% of the cirrhotic patients with low-DNA synthesis activity. Ballardini et found higher reactivity for proliferating cell nuclear antigen in the cirrhotic liver of patients who eventually developed HCC (median: 0.7 vs. 2.61, therefore they suggest a differentiated follow-up of these patients. To our knowledge no data are reported about survival of patients with HCC and proliferation of hepatic cells in the peritumoral cirrhotic liver. Jwo et al.” showed that patients with HCC > 5 cm, or with multipleldaughter HCC nodules had a shortened disease free survival. In the same work patients with diploid or aneuploid pattern but with single GOlGl peak had a reduced overall survival. Other authors showed that patients with aneuploid HCC had a significantly worse prognosis than did those with diploid HCC, also in groups subdivided according to tumor size. Among DNA aneuploid patients, the survival times were significantly less for patients with a low DNA index (<1.5) than for those with a high DNA index ( ~ 1 . 5 ) . ’ By univariate analysis we could not demonstrate a significant correlation of HCC ploidy with survival, whereas we found significant roles for age, number of intrahepatic nodules, Edmondson and Seiner’s classification, portal invasion, vascular invasion, tumor distance from margin of resection, presence of DNA Content and Prognosis in HCC/Rua et al. necrosis, HbsAg, AFP, Child-Pugh’s score, and Sphase fraction of cirrhotic cells. In analogy to our results, a significant association between survival rates and the presence of vascular invasion or intrahepatic metastasis was found in the study of Hamazaki et al.32 and significant association was found between DNA ploidy and HCC prognosis. However, because univariate analysis does not always reflect the actual significance of a prognostic factor, multivariate analysis and calculated cumulative proportion of survival, in relation to clinicopathologic factors, were performed. The Cox’s multivariate survival analysis showed that only Child-Pugh’s score, Sphase fraction of diploid cirrhotic cells, and necrosis were statistically significant independent prognostic factors. Using the same multivariate analysis with Cox’s proportional hazards model, Okada” showed that DNA diploidy, absence of both alcohol abuse and vascular invasion, were independent favorable predictive factors, whereas Fujimoto et al.,’ showed that nuclear DNA content provided significant prognostic value, as did vascular invasion and intrahepatic metastasis. In our study, together with high S-phase fraction of cirrhotic liver parenchyma, Child-Pugh classification was another independent parameter influencing tumor behavior. This can be justified by the impaired function of the liver, after resection of the tumor. In conclusion, our data suggest that S-phase fraction of cirrhotic liver parenchyma may be employed as a new additional parameter in the prognostic evaluation of HCC patients. REFERENCES 1. Tarao K, Ohkawa S, Shimizu A, Harada M, Nakamura Y, Ito Y, et al. Significance of hepatocellular proliferation in the development of hepatocellular carcinoma from anti-hepatitis C virus-positive cirrhotic patients. Cancer 1994;73:114954. 2. Ballardini G, Groff P, Zoli M, Hianchi G, Groff P, Zoli M, Bianchi G, Giostra F, Francesconi €1, et al. Increased risk of hepatocellular carcinoma development in patients with cirrhosis and with high hepatocellular proliferation. J Hepato1 1994;20:218-22. 3. Arakawa M, Kage M, Sugihara S, Nakashima T, Suenaga M, Okuda K. Emergence of malignant lesions within an adenomatous nodule in cirrhotic liver: observation in five cases. Gastroenterology 1988;95:1664-6. 4. Sakamoto M, Hirohashi S, Shimosato Y. Early stages of multistep hepatocarcinogenesis adenomatous hyperplasia and early hepatocellular carcinoma. Hum Puthol 1991; 22: 172-8. 5. Lai CI,, Wu PC, Lam KC, Todd D. Histologic prognostic indicators in hepatocellular carcinoma. Cancer 1979;44:167785. 6. Franco D, Capussotti L, Smadja C, Bouzari H, Meakins J , Kemeny F , et al. Resection of hepatocellular carcinomas. 1201 Results in 72 european patients with cirrhosis. Gastroenterology 1990;98:733-8. 7. Yamanaka N, Okamoto E, Toyosaka A, Mitunobu M, Fujihara S, Kato T, et al. Prognostic factors after hepatectomy for hepatocellular carcinomas. A univariate and multivariate analysis. Cancer 1990;65:1104- 10. 8. Shirabe K, Kanematsu T, Matsumata T, Adachi E, Akazawa K, Sugimachi K. Factors linked to early recurrence of small hepatocellular carcinoma after epatectomy: univariate and multivariate analyses. Hepatology 1991;14:802-5. 9. Fujimoto J, Okamoto E, Yamanaka N, Toyosaka A, Mitsunobu M. Flow cytometric DNA analysis of hepatocellular carcinoma. Cancer 1991;67:939-44. 10. Arii S, Tanaka J, Yamazoe Y, Minematsu S, Morino T, Fujita K, et al. Predictive factors for intrahepatic recurrence of hepatocellular carcinoma after partial hepatectomy. Cancer 1992;69:913-9. 11. Jwo SC, Chiu JH, Chau GY, Loong CC, Lui WY. Risk factors linked to tumor recurrence of human hepatocellular carcinoma after hepatic resection. Hepatology 1992; 16:1367(1. 12. Okada S, Shimada K, Yamamoto J, Takayama T, Kosuge T, Yamasaki S, et al. Predictive factors for postoperative recurrence of hepatocellular carcinoma. Gastroenterology 1994; 106:1618-24. 13. Cong WM, Wu MC. The biopathologic characteristics of DNA content of hepatocellular carcinomas. Cancer 1990; 66~498-501. 14. Anti M, Marra G, Rapaccini GL, Rumi C, Bussa S, Fadda G, et al. DNA ploidy pattern in human chronic liver diseases and hepatic nodular lesions. Flow cytometric analysis on echo-guided needle liver biopsy. Cancer 1994;73: 281-8. 15. Okuda K, Ohtsuki T, Obata H, Tomimatsu M, Okazaki N, Hasegawa H, et al. Natural history of hepatocellular carcinoma and prognosis in relation to treatment. Study of 850 patients. Cancer 1985;56:918-28. 16. Kanai T, Hirohashi S, Upton MP, Noguchi M, Kishi K, Makuuchi M, et al. Pathology of small hepatocellular carcinoma. A proposal for a new gross classification. Cancer 1987; 6018 10- 9. 17. Nakashima S, Okuda K, Kojiro M, Jimi A, Yamaguchi R, Sakamot0 K, et al. Pathology of hepatocellular carcinoma in Japan. 232 consecutive cases autopsied in ten years. Cancer 1983;51:863-77. 18. Edmonson HA, Steiner PE. Primary carcinoma of the liver. A study of 100 cases among 48,900 necropsies. Cancer 1954;71462-503. 19. Stephenson RA, Gay H, Fair WR. Melamed MR. Effect of section thickness on quality of flow cytometric DNA content determinations in paraffin-embedded tissue. Cytometry 1986;7:4 1-4. 20. Coon JS, Landay AL, Weinstein RS. Flow cytometric analysis of paraffin-embedded tumors: implications for diagnostic pathology. Hurn Pnthol 1986;117:435-7. 21. Hedley DW, Friedlander ML, Taylor JW, Taylor IW, Rugg CA, Musgrove EA. Method for analysis of cellular DNA content of paraffin-embedded pathological material using flow cytometry. J Histocliem Cytochein 1983;31:1333-5. 22. Hedley DW, Friedlander ML, Taylor IW. Application of DNA flow cytometry to paraffin-embedded archival material for the study of aneuploidy and its clinical significace. Cytonietry 1985;6:327-33. 1202 CANCER September 15,1996 I Volume 78 I Number 6 23. Joensuu H, Klemi PJ. Comparison of nuclear DNA content in primary and metastatic differentiated thyroid carcinoma. Am J Clin Puthol 1988;89:35-40. 24. Kallioniemi OP, Blanco G, Alavaikko M, Hietanen T, Mattila J, Lauslahti K, et al. Improving the prognostic value of DNA flow cytometry in breast cancer by combining DNA index and S-phase fraction: a proposed classification of DNA histograms in breast cancer. Cancer 1988;62:2183-90. 25. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. JASA 1958;53:457-81. 26. SAS Institute Inc. SAS/STAT User’s Guide. Version 6 4 t h ed., Cary (NC): SAS Institute Inc., 1989. 27. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966;50:163-70. 28. Cox DR. Regression models and life tables. J R Stat SOC 1972;B33: 187-220. 29. SAS Technical Report P-229 SAS/STAT software. Changes and enhancements. Release 6.07. 1992:433-79. 30. Saeter G, Lee CZ, Schwarze E, Ous S, Chen DS, Sung JL, et al. Changes in ploidy distributions in human liver carcinogenesis. JNCI 1988;80:1480-5. 31. Scott RJ, Chakraborty S, Sell S, Hunt JM, Dunsford HA. Change in the ploidy state of rat liver cells during chemical hepatocarcinogenesis and its relationship to the increased expression of a-fetoprotein. Cuncer Res 1989;49:608590. 32. Hamazaki K, Kato T, Yunoki Y, Mori M, Gochi A, Mimura H, et al. An analysis of DNA ploidy pattern of hepatocellular carcinoma. Acta Med Okayumu 1993;47:413-6.