Liver with hypoechoic nodular pattern as a risk factor for hepatocellular carcinoma

GASTROENTEROLOGY1995;108:1778-1784

Liver With Hypoechoic Nodular Pattern as a Risk Factor for Hepatocellular Carcinoma SHINJI KITAMURA,* HIROYASU IISHI,* MASAHARU TATSUTA,t HIDEKI ISHIKAWA,~ TOMOHIKO HIYAMA, § HIDEAKI TSUKUMA, ~ HIROSHI KASUGAI,* SACHIKO TANAKA, II TSUGIO KITAMURA, II and SHINGO ISHIGURO,~ Departments of *Gastroenterology, *Gastrointestinal Oncology, aField Research, IIUItrasonography, and ~Pathology, The Center for Adult Diseases, Osaka, Japan

Background/Aims: Ultrasonography should be used for screening of hepatocellular carcinoma, but there are few reports on the relationship between liver ultrasonographic findings and the development of hepatocellular carcinoma (HCC). Using prospective follow-up studies, we examined the role of liver with a hypoechoic nodular pattern as a high-risk factor in HCC. Methods: The study was performed by follow-up on 593 patients with chronic liver disease recorded at our hospital. The ultrasonographic pattern of the liver parenchyma was classified either as a small or large hypoechoic nodular pattern or as a nonnodular pattern. Patients were followed up from the time of initial ultrasonographic examination ( 1 9 8 5 - 1 9 8 7 ) until January 1, 1991. Results: During the follow-up period (average, 4.2 years, range, 0 . 3 6.0 years), 62 patients were found to have HCC (12%). Patients whose livers showed small or large hypoechoic nodular pattern had a significantly higher risk of HCC than did patients whose livers showed a nonnodular pattern (rate ratios were 14.0 and 20.0, respectively, adjusted for age, sex, hepatitis virus markers, ICG R15, ~-fetoprotein concentration, and ultrasonographic pattern of the liver). Conclusions: Liver showing a hypoechoic nodular pattern is a major risk factor in HCC.

ever, there are few published reports on the relationship between ultrasonographic findings in the liver and the development of HCC. 1° W e conducted a long-term follow-up study on patients with abnormal ultrasonographic findings in the liver to determine whether these abnormal findings constitute a high-risk factor for HCC.

n patients with hepatocellular carcinoma (HCC), early diagnosis offers the only hope for resection and cure. 1 Because this tumor occurs primarily in people with chronic hepatic disease, screening tests have been used to identify small tumors in high-risk patients. 2 Serum hepatitis markers (hepatitis B surface antigen and antibody to hepatitis C virus), Child's A grade, a family history of chronic liver disease and/or hepatitis B virus carrier status, ot-fetoprotein, and natural killer activity of peripheral blood lymphocytes were significant risk factors for the progression of hepatic disease to HCC. 2-6 The most sensitive and specific tests available involve measurement of serum o~-fetoprotein concentrations and high-resolution ultrasonography. 2'7 Ultrasonography is more sensitive than ot-fetoprotein in diagnosing carcinoma and should be used for screening purposes, s'9 H o w -

All patients were examined by abdominal ultrasonography when they were enrolled. A high-resolution, real-time scanner (Hitachi EUB-40 and -340; Hitachi, Tokyo, Japan) was used, equipped with a 3.5-MHz linear probe that provides a dynamic evaluation of abdominal structures. The ultrasonographic images were recorded on instant films. Based on the size and distribution of liver hypoechoic nodules, echo patterns of the liver parenchyma were classified into the following four patterns u on the initial ultrasonograms by one physician (S.K.) without knowledge of clinical data: (1) homogeneous echoic pattern, no hypoechoic nodular pattern (Figure 1A); (2) coarse echoic pattern, liver parenchyma showing coarse echoic pattern but no distinct hypoechoic nodules (Figure 1B); (3) small hypoechoic nodular pattern, echo pattern

I

P a t i e n t s and M e t h o d s Patient Enrollment We examined the role of liver with a hypoechoic nodular pattern as a high-risk factor in HCC prospectively. We enrolled in this study a total of 593 patients examined by abdominal ultrasonography. All met the following criteria: (1) patients living in Osaka prefecture, (2) a clinical diagnosis of chronic hepatitis or compensated cirrhosis (without ascites, jaundice, or a history of bleeding esophageal varices) during the period of January 1985 to December 1987, (3) no history of liver cancer or evidence of this tumor at the time of enrollment (patients in whom liver cancer was detected at the initial examination or diagnosed within 3 months after enrollment were excluded, and (4) no other serious diseases.

Classification of Ultrasonographic Pattern of the Liver

© 1995 by the AmericanGastroenterologicalAssociation

0016-5085/95/$3.00

June 1995

HYPOECHOIC NODULAR PATTERN AND HCC

1779

Figure 1. (A) Homogeneous echoic pattern. No hypoechoic nodules could be observed in the liver. (B) Coarse echoic pattern. Ultrasonographic pattern of the liver parenchyma was coarse, but no hypoechoic nodules could be observed in the liver. (C) Small hypoechoic nodular pattern. Ultrasonographically, many hypoechoic nodules 3 - 5 mm in diameter were observed in the liver. (D) Large hypoechoic nodular pattern. Ultrasonographically, many hypoechoic nodules ->5 mm in diameter were observed in the liver.

showing scattered hypoechoic nodules as large as 3 - 5 mm in diameter (Figure 1C); (4) large hypoechoic nodular pattern, echo pattern showing scattered hypoechoic nodules -->5 mm in diameter (Figure 1D). A few hyperechoic nodules were found in this series, but these nodules were excluded from this analysis.

Follow-up All patients were followed up until January 1, 1991. All available clinical follow-up data were first collected, including those on the last visiting date, (x-fetoprotdn, ultrasonographic examination, and computed tomography. Second, patients were surveyed by telephone for vital status. Third, the patients were followed up by record linkage with the Osaka Cancer Registry. 3':2 All patients were observed from the time of their first ultrasonographic examination until HCC was diagnosed or until the end of the study period (January 1, 1991).

Diagnosis of HCC Space-occupying lesions detected or suspected at the time of ultrasonographic examination or 0¢-fetoprotein mea-

surement were further examined with computed tomography, hepatic arteriography, fine-needle aspiration biopsy, and other such clinical diagnostic techniques as hepatic scintigraphy unless the ultrasonographic findings confirmed lesions to be unquestionably benign (e.g., hemangioma or cysts). Space-occupying lesions was defined as localized lesions with clear margins more than 1 cm in diameter. These lesions were ultrasonographically suspected of HCC. A final diagnosis of HCC was based on histological specimens or on the radiological findings of hepatic arteriography.

Labeling Index of Biopsy Specimens in Relation to Ultrasonographic Findings in the Liver The proliferation activity of nontumorous hepatocytes was examined in 16 patients with space-occupying lesions in the liver. Liver specimens were obtained from the tumor and nontumorous regions using a fine-needle aspiration biopsy under ultrasonographic guidance. Histological examinations of the liver specimens showed that lesion was benign in all patients. Before the biopsies were performed, informed consent was obtained from all patients.

1780 KITAMURAETAL.

Hepatocytic proliferation activity was measured by DNA polymerase-O~ immunoreactivity. 13 The fresh liver specimens were fixed in a mixture of periodate, lysine, and 2 % paraformaldehyde; embedded in OCT compound (Miles Scientific, Naperville, IL); and frozen in liquid nitrogen. Before staining, frozen sections (6-~m-thick) were made and washed in 0.01 mol/L phosphate-buffered saline (PBS). The sections were then treated for 20 minutes at room temperature with 10% normal goat serum, covered with a drop of a solution of mouse monoclonal antibody against DNA polymerase 0~(Medical & Biological Laboratories, Nagoya, Japan) or of normal mouse serum without mouse monoclonal antibody against DNA polymerase o~ (control), and left overnight at 4°C. To inhibit endogeneous enzyme activity, the sections were then incubated with methyl alcohol containing 0.03% hydroperoxidase, rinsed with PBS five times for 5 minutes each time, and treated for 6 hours with goat polyclonal antibody against mouse immunoglobulin G at 4°C. A peroxidase-antiperoxidase complex with goat antibody was used, and the sections were stained with 3,3'-diaminobenzidine tetrahydrochloride with 0.03% hydroperoxide and washed for 10 minutes in PBS. The sections were counterstained for nuclei with 10% methyl green and mounted with mounting medium. Nuclei of cells that reacted with the monoclonal a n t i - D N A polymerase O~antibody were brown by light microscopic examination. To analyze the labeling index of hepatocytes, the number of DNA polymerase-o~-labeled and -unlabeled cells among 200-300 ceils was counted. Investigators performing the analysis did not know from which groups the samples had come. The labeling index was calculated as number of labeled ceils per total number of ceils examined.

Statistical Analysis The method of Kaplan and Meier 14 was used to estimate the cumulative risk of HCC according to the echo pattern of the liver at the time of enrollment. Cox proportional-hazards regression analysis I5 was performed to estimate the rate ratios of possible risk factors for HCC. The period of observation used in calculating the risk of HCC began at the date of enrollment and ended when liver cancer was diagnosed, when the patient died, or on January 1, 1991, whichever came first. Data were analyzed using the STATA statistical package (Computing Resource Center, Santa Monica, CA). The level of significance was calculated to be P < 0.03.

Results Characteristics of Patients at Time of Enrollment The characteristics of patients with chronic liver disease in relation to the ultrasonographic pattern of the liver at the time of enrollment are summarized in Table 1. A total of 593 patients met the study criteria and were registered; 205 were clinically diagnosed with liver cirrhosis; and 388 had chronic hepatitis. One hundred three patients with liver cirrhosis (50%) had hypoechoic

GASTROENTEROLOGYVol. 108, No. 6

nodular patterns, but the others did not have hypoechoic nodular patterns. The proportion of patients 50 years of age or younger was significantly smaller among those with small or large hypoechoic nodular patterns in the liver than among those with homogeneous echoic pattern. Hepatitis B surface antigen was detected significantly more frequently in patients with small or large hypoechoic nodular patterns. Table 1 also shows that values of I C G R15, total bilirubin, and serum o~-fetoprotein were significantly higher, but serum albumin and total cholesterol levels were significantly lower, in patients with these nodular patterns.

Follow-up The numbers of patients with H C C and various ultrasonographic patterns are shown in Table 2. The vital status of 30 patients was unknown at the end of the study period, resulting in a follow-up rate of 94.9%. During the observation period (from enrollment to January 1, 1991; mean + SE duration, 4.2 -+- 0.1 years; range, 0 . 3 - 6 , 0 years), 62 patients (12%) were found to have HCC. O f these patients, 52 were found to have liver cirrhosis at the time of enrollment, and 10 were found to have chronic hepatitis. Table 3 summarizes the results of changes in ultrasonographic pictures during the follow-up period (average interval, 1 year; range, 3 months to 2 years). Two hundred thirty-two (73 %) of 318 ultrasonographic pictures remained constant, and ultrasonographic changes were found in 86 (27%) with time.

Cumulative Risk of HCC The Kaplan-Meier estimates of the cumulative risk of HCC in relation to the ultrasonographic liver findings at the time of enrollment are shown in Figure 2. The 6year cumulative risks of HCC (+SE) were 1.5% (+0.9%) and 7.7% (___2.0%) in patients with homogeneous and coarse echoic patterns, respectively, and 39.9% (+6.3%) and 54.7% (+11.7%), respectively, in patients with small and large hypoechoic nodular patterns. A log-rank test of the four curves showed significant differences (P < 0.05).

Hazard Rate Ratios for HCC Table 4 presents the results of Cox proportional hazards regression analyses in which sex, age, and other possible confounders (hepatitis B surface antigen, antibody to hepatitis C virus, I C G R15 levels, serum O~fetoprotein concentrations, and ultrasonographic liver patterns) were adjusted to simultaneously estimate hazard rate ratios for developing HCC. The risk of developing H C C was significantly higher in men than in women with the risk 4.6 times greater

June 1995

HYPOECHOIC NODULAR PATTERN AND HCC

1781

Table 1. Characteristics of Patients With Chronic Liver Disease in Relation to the Ultrasonographic Pattern of the Liver

Characteristic No. Sex (%) Male Female Age (%)a 2 0 - 4 9 yr 5 0 - 5 9 yr 6 0 - 8 3 yr Mean + SE Hepatitis marker (%) Hepatitis B surface antigen" Positive Negative Unknown Antibody to hepatitis C virus Positive Negative Unknown ICG R15 (%)° Serum alanine aminotransferase ( U / L ) c Total bilirubin ( m g / d L ) ° Serum albumin ( g / d L ) c Serum T-glutamyltranspeptidase ( U / L ) c Total cholesterol ( m g / d L ) ° Serum G-fetoprotein ( n g / m L ) c Stage of disease (%)8 Liver cirrhosis Chronic hepatitis

Patients with homogeneous echoic pattern

Patients with coarse echoic pattern

232

231

143 (62) 89 (38)

129 (56) 102 (44)

91 (39) 77 (34) 64 (27) 52 ± 1

61 95 75 54±

Patients with small hypoechoic nodular pattern

Patients with large hypoechoic nodular pattern

100

30

58 (58) 42 (42)

24 (80) 6 (20)

(26) (41) (33) 1

16 (16) 41 (41) 43 (43) 58 ± i °

6 (20) 8 (27) 16 (53) 58_+ 2 °

5 (2) 206 (89) 21 (9)

16 (7) 199 (86) 16 (7)

9 (9) 86 (86) 5 (5)

4 (13) 25 (84) 1 (3)

22 (9) 29 (13) 181 (78) 14.2 ± 1.1 93 ± 7 0.77 + 0.02 4.4 _+ 0.2 90 ± 9 203 ± 3 20 ± 16

41 (18) 32 (14) 158 (68) 21.2 + 1.2 105 + 7 0,90 + 0.03 4.2 + 0.4 80 _+ 7 190 ± 3 55 + 16

15 (15) 5 (5) 80 (80) 30.8 + 2.2 ° 117 _+ 11 1.19 ± 0.07 ° 3.9 _+ 0.5 ° 115 _+ 15 184 _+ 5 ° 76 ± 32 °

4 (13) 2 (7) 24 (80) 27.7 ± 3.5 ° 123 ± 20 1.13 ± 0.09 b 4.0 ± 0.9 ° 122 ± 20 183 + 7 ° 269 ± 169 b

17 (7) 215 (93)

85 (37) 146 (63)

80 (80) 20 (20)

23 (77) 7 (23)

"A significant correlation was found among the four groups (P < 0.05). °Significantly different from the value for patients with homogeneous echoic patterns (P < 0.05). CMean _+ SE.

in men. A positive association was observed between the risk of liver cancer and age at the time of enrollment; patients in their 50s and 60s had significantly higher rate ratios (3.0 and 2.9, respectively) than did those younger than 49 years of age. The serum oc-fetoprotein concentration at the time of enrollment was also found to be a significant risk marker for HCC. Table 3 also shows that such associations were less evident between serum markers for hepatitis viruses (hepatitis B surface antigen and antibody to hepatitis C virus) and ICG R15 values and the risk of liver cancer. Ultrasonographic liver patterns were significantly as-

sociated with the risk of HCC. The adjusted rate ratios for liver with small and large hypoechoic nodular patterns were estimated to be 14.0 and 20.0, respectively. This finding indicated that nodular-patterned liver is a major risk factor for HCC.

Histological Features of the Biopsy Specimens and Labeling Index of Hepatocytes in Relation to Ultrasonographic Pattern in the Liver Table 5 summarizes data on histological features of the biopsy specimens and the labeling index of nontu-

Table 2. Observed Numbers of HCCs in Relation to the Ultrasonographic Pattern of the Liver

No. Average observation period (yr) a Patients found to have HCC (%) aMean _+ SE.

Patients with homogeneous echoic pattern

Patients with coarse echoic pattern

Patients with small hypoechoic nodular pattern

Patients with large hypoechoic nodular pattern

Total

232 4.3 ± 0,1 3 (1)

231 4.3 + 0,1 15 (7)

100 3.6 _+ 0.2 31 (31)

30 3.3 + 0.4 13 (43)

593 4.2 ± 0.1 62 (12)

1782

KITAMURA ET AL.

GASTROENTEROLOGY Vol. 108, No. 6

Table 3. Changes of Ultr~,sonographic Pictures During the Follow-up Period Echo patterns at the initial examinations (%) Homogeneous echoic pattern Echo patterns at follow-up examinations Homogeneous echoic pattern Coarse echoic pattern Small hypoechoic nodular pattern Large hypoechoic nodular pattern Total

77 (71) 31 (29)

108 (100)

morous hepatocytes in relation to the ultrasonographic pattern in the liver. Histological examination of biopsy specimens showed that adenomatous hyperplasia was found more frequently in specimens obtained from patients with small or large hypoechoic nodular patterns than in those from patients with homogeneous or coarse echoic pattern. No HCCs were found in the biopsy specimens obtained from patients with nodular patterns. Patients whose livers showed small or large nodular patterns had a significantly greater labeling index of the hepatocytes than did patients whose livers showed homogeneous or coarse echoic patterns.

Discussion In the present study, we found that a liver with a hypoechoic nodular pattern presents a major risk for HCC. During the follow-up period ending January 1, 1991, 62 patients had this form of cancer diagnosed. The risk of developing it was significantly increased in patients with hypoechoic nodular patterns in the liver; the rate ratios for patients with small and large nodular patterns were 14.0 and 20.0, respectively. Mikami et al. reported similar results. 1° Making special reference to ultrasonographic findings of the liver parenchyma, they conducted a follow-up study of 0.5-3.5 years' duration to the development of HCC in 90 patients with liver

100

I

v Large

nodular

Q:

50 ¸

E

Coarse

(O J

J

1

,

,

2

,

...... ,

3

~"-""~"~-'~ ,

,

4

,

,

5

H o m o g e n e o u s

,

6

Years of Follow-up Figure 2. Cumulative risk of HCC in 593 patients with chronic liver

disease in relation to the ultrasonographic pattern in the liver. A logrank test of the four curves showed significant differences (P < 0.05).

Coarse echoic pattern

Small hypoechoic nodular pattern

18 (13) 100 (74) 18 (13) 136 (100)

6 43 7 56

Large hypoechoic nodular pattern

(10) (77) (13) (100)

6 (33) 12 (67) 18 (100)

cirrhosis. They found that HCC developed more frequently in patients with hypoechoic nodular patterns than in those without them. 1° The reason for association between hypoechoic nodular patterns in the liver and this type of cancer is unknown, but at least two possibilities may be considered. One is an effect of the macronodular type of liver cirrhosis on hepatocarcinogenesis. Histologically, the size of regenerative nodules in the liver plays an important role in the

Table 4. Results of Cox Proportional-Hazards Regression,

Adjusted for Age, Sex, and Other Potential Confounders

Variable Sex Female Male Age (yr) 19-49 50-59 60-83 Hepatitis B surface antigen Negative Positive Antibody to hepatitis C virus Negative Positive ICG R15 (%) 0-12.9 13.0-23.3 ->23.4 ~-Fetoprotein (ng/mL) 0-19 20-99 ->100 Ultrasonographic pattern Homogeneous echoic pattern Coarse echoic pattern Small hypoechoic nodular pattern Large hypoechoic nodular pattern

Adjusted rate ratio

95% confidence interval

P

1 4.6

2,2-9.5

<0.01

1 3.0 2.9

1.3-7.2 1.2-7.0

0.01 0.02

1 1.2

0.5-2.8

0.63

1 1.6

0.3-3.9

0.82

1 0.8 1.5

0.3-2.5 0.6-4.3

0.69 0.42

1 3.1 4.0

1.4-6.8 1.8-8.8

0.01 <0.01

1 3.2

0.9-11.5

0.07

13.9

3.9-49.3

<0.01

20.0

5.3-76.0

<0.01

June 1995

HYPOECHOIC NODULAR PATTERN AND HCC

Table B, Labeling Index of N o n t u m o r o u s H e p a t o c y t e s and the Presence of A d e n o m a t o u s Hyperplasia in Relation to t h e Ultrasonographic Pattern in the Liver

Homogeneous or coarse echoic pattern No. of specimens examined Adenomatous hyperplasia (%) Positive Negative Labeling index (%)

7 1 (14) 6 (86) 2.8 _+ 1.3

Small or large hypoechoic nodular pattern 9 5 (56) 4 (44) 4.2 _+ 1.4 a

aSignificantly different from the value for patients with homogeneous or coarse echoic pattern in the liver at P < 0.05.

incidence of HCC. ~6'17 Kew pointed out that the greater risk of HCC in the presence of cirrhosis in Africa and southeast Asia is associated with a macronodular pattern. 16 This close relationship also has been reported by Nonomura et al. among Japanese patients 18'.9 and by Patterson et al. in the black population of Southern Africa. 2° Nakamura et al. also examined nonneoplastic morphological changes in various types of liver cirrhosis in relationship to the presence or absence of HCC in Japan and concluded that patients with macronodular cirrhosis may have an increased risk of developing the cancer. 21 However, in our study, histological diagnosis of macronodular cirrhosis by biopsy specimens was very difficult. Kimura et al. found that regenerative nodules in the liver could be seen as hypoechoic nodules by ultrasonography and that macronodules were more frequently found histologically in liver showing a hypoechoic nodular pattern than in that showing no such pattern. 1~ Therefore, these findings suggest that the liver showing this pattern is at greater risk for HCC. Another possible explanation for this association is the effect of accelerated cell proliferation on hepatocarcinogenesis. Increased mitotic activity of tissue cells correlates with the development of cancers, presumably by an increased rate of random mutation 22 and by promotion. 23 There also have been many hepatocarcinogenesis experiments conducted in animals, and their results support the above-mentioned hypothesis. Tarao et al. prospectively followed the change in DNA synthesis activity in hepatocytes from cirrhotic patients and its relationship to the development of HCC and found that patients with persistently high DNA synthesis activity or increased activity are at very high risk for tumor development. During their 2-year follow-up period, HCC developed in 1 1 of 33 patients with liver cirrhosis (8 of 15 patients with

1783

high DNA synthesis activities initially and 3 of 18 with . . . . . . ,mtlally). 24 In the present work, we found low act,wtles that patients whose livers showed hypoechoic nodular patterns had significantly higher DNA polymerase-0~ labeling index than did patients without such patterns. This study indicates that livers with hypoechoic nodular patterns present a major risk for HCC. These results suggest that more careful and frequent examinations with various imaging modalities are needed for early detection of HCC in patients with such liver patterning.

References 1. Ellis JC. Screening for hepatocellular carcinoma. Review and perspective. West J Med 1 9 8 8 ; 1 4 9 : 1 8 3 - 1 8 7 . 2. Regan LS. Screening for hepatocellular carcinoma in high-risk individuals. A clinical review. Arch Intern Med 1989; 1 4 9 : 1 7 4 1 1744. 3. Tsukuma H, Hiyama T, Tanaka S, Nakao M, Yabuuchi T, Kitamura T, Nakanishi K, Fujimoto I, Inoue A, Yamazaki H, Kawashima T. Risk factor for hepatocellular carcinoma among patients with chronic liver disease. N Engl J Med 1 9 9 3 ; 3 2 8 : 1 7 9 7 - 1 8 0 1 . 4. Kobayashi K, Unoura M, Tanaka N, Hattori N. A comparison between hepatocellular carcinoma-developing and non-carcinomadeveloping patients with cirrhosis over a long follow-up period. Hepatogastroenterology 1 9 9 0 ; 3 7 : 4 4 5 - 4 4 8 . 5. DiBiscelgie AM, Rustgi VK, Hoofnagle JH, Dusheiks GU, Lotze MT. Hepatocellular carcinoma. Ann Intern Med 1988; 1 0 8 : 3 9 0 401. 6. Nakajima T, Mizushima N, Kanai K. Relationship between natural killer activity and development of hepatocellular carcinoma in patients with cirrhosis of the liver. Jpn J Clin Oncol 1987; 17:327-332. 7. Tanaka S, Kitamura T, Nakanishi K, Okuda S, Kojima J, Fujimoto I. Recent advances in ultrasonographic diagnosis of hepatocellular carcinoma. Cancer 1 9 8 9 ; 6 3 : 1 3 1 3 - 1 3 1 7 . 8. Sheu JC, Sung JL, Cheu DS, Lai MY, Wang TH, Yu JY, Yang PM, Chuang CN, Yang PC, Lee CS, Hsu HC, How SW. Early detection of hepatocellular carcinoma by real-time ultrasonography. Cancer 1985;56:660-666. 9. Cottone M, Turri M, Caltagirone M, Maringhini A, Sciarrino E, Virdone R, Fusco G, Orlando A, Marino L, Pagliaro L. Early detection of hepatocellular carcinoma associated with prospective study. Hepatogastroenterology 1 9 8 8 ; 3 5 : 1 0 1 - 1 0 3 . 10. Mikami N, Ebara M, Yoshikawa M, Ohto M. Relationship between ultrasound-findings of low-echoic nodule of hepatic parenchyma in liver cirrhosis and development of hepatocellular carcinoma. Jpn J Gastroenterol 1 9 9 0 ; 8 7 : 1 0 1 0 - 1 0 1 9 . 11. Kimura T, Ebara M, Ohto M, Kondo F. Classification of liver cirrhosis based on parenchymal echo patterns and its clinical usefulness for diagnosis of liver cirrhosis. Jpn J Gastroenterol 1989; 8 6 : 1 4 7 3 - 1 4 8 5 . 12. Oshima A, Sakagami F, Hanai A, Fujimoto I. A method of recording linkage. Environ Health Perspect 1 9 7 9 ; 3 2 : 2 2 1 - 2 3 0 . 13. Seki S, Sakaguchi H, Kawakita N, Yanai A, Kuroki T, Mizoguchi Y, Kobayashi K. An analysis of proliferating cells in biopsy specimens from patients with small hepatocellular carcinoma. Cancer 1992; 6 9 : 2 4 3 3 - 2 4 3 9 . 14. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1 9 5 8 ; 5 3 : 4 5 7 - 4 8 1 . 15. Cox DR. Regression models and life-tables. J R Stat Soc 1972;34:187-220. 16. Kew MC. Clinical, pathologic and etiologic heterogeneity in hepatocellular carcinoma: evidence from Southern Africa. Hepatology 1981; 1 : 3 6 6 - 3 6 9 .

1784

KITAMURA ET AL.

17. Popper H. Pathologic aspects of cirrhosis. A review. Am J Pathol 1977; 7:228-264. 18. Nonomura A, Hayashi M, Watanabe K, Takayanagi N, Ohta G. Study on the pathogenesis of hepatocelluiar carcinoma in HBVnegative alcoholic cirrhosis. Acta Pathol Jpn 1986;36:12971350. 19. Nonomura A, Hayashi M, Takayanagi N, Watanabe K, Ohta G. Correlation of morphologic subtypes of liver cirrhosis with excess alcohol intake, HBV infection, age at death, and hepatocellular carcinoma. Acta Pathol Jpn 1986;36:631-640. 20. Patterson AC, Kew MC, Herman AB, Becker PJ, Hadkinson J, Isaacson C. Liver morphology in southern Africa blacks with hepatocellular carcinoma: a study within the urban environment. Hepatology 1984; 5:72-78. 21. Nakamura Y, Kanel GC, Doishita K, Ohta G, Peters RL. Risk lesions in cirrhosis and development of hepatocellular carci-

GASTROENTEROLOGY Vol. 108, No. 6

noma: an autopsy study. Biomed Pharmacother 1989;43:127133. 22. Slaga TJ, Scribner JD. Inhibition of tumor initiation and promotion by antiinflammatory agents. JNCl 1973; 51:1723-1725. 23. Argyris TS. Tumor promotion by damage-induced epidermal hyperplasia in the skin of mice (abstr). Proc Am Assoc Cancer Res 1979;20:4. 24. Tarao K, Shimizu A, Ohkawa S, Harada M, Ito Y, Tamai S, Kuni Y, Okamoto N, Inoue T, Kanisawa M. Development of hepatocellular carcinoma associated with increases in DNA synthesis in the surrounding cirrhosis. Gastroenterology 1992; 103:595-600.

Received May 4, 1994. Accepted February 1, 1995. Address requests for reprints to: Shinji Kitamura, M.D., Second Department of Internal Medicine, Osaka University Medical School, 2-2, Yamadaoka, Suita-city, Osaka 565, Japan. Fax: (81) 6-8793739.