(CANCER RESEARCH 51, 2842-2847. June 1. 1991]
Hepatitis C and Hepatitis B in the Etiology of Hepatocellular Carcinoma in the
Japanese Population1
Keitaro Tanaka,2 Tornio Hirohata, Shunichi Koga, Keizo Sugimachi, Takashi Kanematsu, Fumitake Ohryohji,
Hajimc Nawata, Hiromi Ishibashi, Yoshiaki Maeda, Hiroyuki Kiyokawa, Kazuo Tokunaga, Yoshiko Irita,
Setsuko Takeshita, Yasuko Arase, and Noriko Nishino
Department of Public Health [K. Tu., T. II.], the First Department [H. I.j and the Third Department fH. N.¡of Internal Medicine, and the Second Department of Surgery
[K. S„T. A'./, Kyushu University School oj Medicine, 3-I-I Maidashi, Higashi-ku, Fukuoka 812; Ihuka Hospital, 3-83 Yoshiomachi, li:uka-shi 820 [S. K]; Health
Promotion Section, Bureau of Public Health oj Fukuoka City, 1-8-1 Tenjin, Chuo-ku, Fukuoka 810 ¡F.O.¡:Fukuoka Red Cross Blood Center, 232-11 Oa:akamikoga,
Chikushino-shi 818 [Y. A/., //. A., A'. To., Y. 1.];Department oj Information and Management Science, Tokai ( niversily Fukuoka Junior College, 1137 Taku, Munakatashi 811-41 [S. T.I; and Hakata Public Health Center, 1-21-16 Chiyo, llakala-ku, Fukuoka 812 fY. A., ,V. NJ. Japan
ABSTRACT
We conducted case-control studies of hepatocellular carcinoma (HCC)
and liver cirrhosis (l.( ) in relation to hepatitis C virus (HCV) and
hepatitis B virus infection, involving 91 patients with HCC, 75 patients
with LC who had no evidence of HCC, and 410 control subjects from the
Japanese population. Serum antibody to HCV (anti-HCV) was detected
by both enzyme-linked immunosorbent assay and recombinant immunoblot assay in 51, 51, and 3% of HCC, LC, and controls, respectively,
whereas the corresponding prevalence of serum hepatitis B surface
antigen (HBsAg) was 21, 11, and 2%, respectively. The relative risks
(and 95% confidence intervals) for the presence of serum anti-HCV were
estimated as 52.3 (23.9-114.3) for HCC and 64.4 (27.4-151.4) for LC.
These values exceeded the relative risk of HCC (15.3) and that of LC
(6.1) for positive serum HBsAg. Among male patients with HCC or LC,
anti-HCV rates were very high in blood recipients (about 70%), heavy
drinkers (46-62%), and those who had no identifiable risk factors (6575%), indicating possible transmission of HCV via routes other than
transfusion. No significant difference in anti-HCV status was observed
between the HCC and LC groups. It was notable that anti-HCV was
much less prevalent among HBsAg-positive patients with HCC or LC
than among HBsAg-negative ones. There was a slight to moderate
increase in HCC or LC risk among blood recipients and heavy drinkers
after adjustment for anti-HCV status. These results indicate that, in
Japan, the possible role of HCV infection in the etiology of HCC and
LC is extremely large and seems to be more important than chronic
hepatitis B virus infection.
Hence, risk factors other than chronic HBV infection, especially
infection with unidentified non-A, non-B hepatitis virus or
viruses, have been suspected to be important in the etiology of
HCC in Japan (10-12).
In 1989, Choo et al. (13) reported the discovery of comple
mentary DNA presumed to be derived from a blood-borne nonA, non-B hepatitis virus. They designated the virus as HCV,
and now an assay for anti-HCV has become available (14). We
have conducted case-control studies of HCC and LC especially
focusing on the role of HCV and HBV infection in Fukuoka
prefecture, where the risk for HCC is one of the highest in
Japan (15). In this area, liver cancer is the second and third
leading cause of cancer deaths in men and women, respectively
(crude death rate per 100,000 of liver cancer in 1987: men,
47.8; women, 15.9).
MATERIALS
AND METHODS
Study Subjects. We initiated a case-control study of HCC in relation
to various possible risk factors in 1985. In addition to determining
serum HBV markers, a detailed interview survey on past histories of
blood transfusion, alcohol consumption, cigarette smoking, etc., was
conducted in the study. The methods and part of the results obtained
have been reported previously (12). In brief, patients with HCC were
eligible if the following criteria were met: (a) diagnosed as HCC initially
within 1 year prior to identification; (b) ages 40 to 69 years (patients
ages 70 years or over as well as those under 40 years were excluded
INTRODUCTION
since the former may lack reliability in their responses about their past
To date, the most important risk factor for HCC' has been histories and the latter presented a practical difficulty in obtaining
adequate controls with respect to age); (c) residents in Fukuoka or Saga
chronic HBV infection (1-3). In fact, a great majority of HCC
Prefecture (neighboring Fukuoka prefecture) of Japanese nationality.
patients in the regions endemic for the disease such as sub- Patients with HCC were identified among those who were admitted to
Saharan Africa and southeast Asia seem to be attributed to the First and Third Departments of Internal Medicine and the Second
chronic HBV infection (4, 5). Japan is also among the relatively
Department of Surgery of Kyushu University Hospital from December
1985 to June 1989 by contact with physicians as well as by a weekly
high risk areas (6). However, in this country, most HCC pa
review of the list of hospitalized patients and the admission reports.
tients (about 75%) in recent years have been negative for serum
Finally, a total of 204 Japanese patients with HCC who met the above
HBsAg (7), a marker which, when testing positive, usually
indicates a chronic HBV carrier state. Although Bréchotet al. criteria were investigated. Among the 204 patients, the sera from 91
(8) reported that HBV DNA was frequently detected in liver patients were stored and available for this study. These 91 patients did
tissue from HBsAg-negative patients with HCC and thus HBV not significantly differ from the remaining 113 patients without their
stored sera in the prevalence of serum HBsAg and histories of blood
may play a role in those patients, such evidence has not been transfusion and alcohol consumption. Eighty-nine of the 91 patients
confirmed in studies using the same approach in Japan (9, 10). were residents of Fukuoka prefecture, while 2 were from Saga prefec
ture. The major diagnostic methods for HCC in the 91 patients were:
Received 9/17/90; accepted 3/18/91.
30 histologically confirmed; 58 based on angiographie findings (16); 3
The cosls of publication of this article were defrayed in part b> the payment
based on the findings of ultrasonography and computed tomography,
of page charges. This article must therefore be hereby marked advertisement in
accordance with 18 ll.S.C. Section 1734 solely to indicate this fact.
elevated serum «-fetoprotein. etc. Preexisting liver diseases among the
' This work was supported in part by grants from the Ministry of Health and
91 patients were: 76 with liver cirrhosis (25 histologically confirmed; 1
Welfare, the Ministry of Education. Science and Culture (60480195). Japan, and
confirmed by laparoscopy; 50 clinically established); 2 with chronic
the Fukuoka Cancer Society.
2 To whom requests for reprints should be addressed.
active hepatitis; 3 with chronic persistent hepatitis; 4 with chronic
3The abbreviations used are: HCC, hepatocellular carcinoma: HBV. hepatitis
hepatitis without available histology; 1 with no diagnostic abnormality;
B virus: HBsAg, hepatitis B surface antigen: HCV, hepatitis C virus: anti-HCV.
5 unknown. Of the 91 patients: 19 (21%) were serum HBsAg positive;
antibody to HCV: LC. liver cirrhosis; ELISA, enzyme-linked immunosorbent
22 (24%) had positive history of blood transfusion due to causes other
assay: RIBA, recombinant ¡mmunoblotassay; SOD. Superoxide dismutase: RR.
than liver diseases; 27 (30%) had positive history of heavy drinking (80
relative risk; CI, confidence interval: PAR, population attributable risk.
2842
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
HEPATITIS C AND HEPATOCELLULAR CARCINOMA
ml or more of ethanol per day for at least 10 years); 34 (37%) had no
identifiable risk factors; 7 patients had double factors (2 with positive
HBsAg and transfusion; I with positive HBsAg and heavy drinking; 4
with positive transfusion and heavy drinking); and 2 patients had triple
factors.
In parallel with the case-control study. 100 Japanese patients with
LC were also investigated at the same departments during the period
from December 1985 through December 1987 according to the same
methods and by the same interviewers. Patients with liver cirrhosis who
met the above criteria of A and c and who had no evidence of HCC by
available diagnostic methods such as ultrasonography, computed to
mography, and serum n-fetoprotein at the time of identification were
regarded as eligible. Special types of LC such as primary and secondary
biliary cirrhosis and that due to autoimmune hepatitis, parasitosis,
congestive heart failure, or metabolic disorders were excluded because
they appeared to differ substantially from the most common type of
LC observed in Japan. Because of a possible bias in hospitalized patients
with LC with respect to relevant factors, most (67%) of the patients
were selected from among the outpatients. Among the 100 patients,
the sera from 75 patients were available for this study. Seventy-two of
these 75 patients were residents of Fukuoka prefecture. The diagnoses
of LC for the 75 patients were based on the following methods: 24
histologically confirmed; 14 confirmed by laparoscopy; 37 based on
evident clinical signs and the findings in imaging methods such as
ultrasonography (17) and laboratory data. Of the 75 patients: 8 (11%)
were serum HBsAg positive; 21 (28%) had positive history of blood
transfusion due to causes other than liver diseases; 14 (19%) had
positive history of heavy drinking; 41 (55%) had no identifiable risk
factors; and 9 patients had double factors (3 with positive HBsAg and
transfusion, and 6 with positive transfusion and heavy drinking).
Controls comprised 410 persons ages 40 to 69 years who were
residents of Fukuoka City and who v¡siteda public health center located
near Kyushu University Hospital between January 1986 and July 1989
to undergo health examinations. The controls were selected so that
their distribution by sex and age was as similar as possible to that of
patients with HCC or LC. Those who suffered from chronic liver
diseases such as chronic hepatitis and liver cirrhosis, or from whom
blood specimens could not be obtained, were excluded. Sera from all
the controls were stored and utilized for this study. The distribution of
study subjects by demographic characteristics is shown in Table 1. As
for occupation, subjects were classified according to the jobs in which
they had been engaged longest. No patients with HCC or LC had
experienced occupational exposure to special chemical substances.
There were some differences among study groups in the following
points: the age between male HCCs and male controls (P < 0.05 by
VV'ilcoxontest); occupations between male HCCs and male controls
and between male LCs and male controls (P < 0.05 by x2 test); years
of education between female HCCs and female controls and between
female LCs and female controls (P < 0.01 by Wilcoxon test). These
factors were accommodated by the statistical procedures in estimating
the relative risks for relevant factors because of their possible confound
ing effects.
Interviews. Concerning alcohol consumption, history of blood trans
fusion, and other relevant factors in addition to demographic charac
teristics, all study subjects were interviewed in person by three trained
interviewers. During the first 2.5 years, one person interviewed both
patients and controls, and over the next 2 months another interviewer
conducted interviews while taking turns with the first interviewer due
to logistic reasons, and the last interviewer conducted all interviews for
the rest of the study period. Most of the subjects were interviewed by
the first interviewer. To obtain accurate information, all interviews were
tape recorded and checked against questionnaires by both the interview
ers and one of the authors (K. Tanaka). Subjects were interviewed in a
private room in the hospital wards or in the outpatient clinic or at the
health center.
Laboratory Tests. Information on serum HBV markers was obtained
from medical records for patients: all patients were tested for serum
HBsAg by radioimmunoassay or reversed passive hemagglutination;
the status of antibody to hepatitis B core antigen or antibody to HBsAg
was not determined in most of the patients. For the controls, sera were
tested for HBsAg by the reversed passive hemagglutination method.
Sera from 91 patients with HCC, 75 patients with LC. and 410 controls
were kept frozen at -70°C in the same deep freezer until tested for
anti-HCV. The Ortho-HCV ELISA (Ortho Diagnostic Systems, Raritan, NJ) was used to detect anti-HCV. This indirect assay uses a
recombinant HCV antigen polypeptide (C 100-3). described in detail by
Choo et al. (13). Anti-HCV in the serum binds solid-phase HCV
antigen, and the specifically captured antibody is revealed with an
antibody-enzyme conjugate composed of a murine monoclonal antibody
against human IgG and the enzyme, conjugated with horseradish peroxidase. The test was performed according to the manufacturer's in
structions. The results were read at 492 nm by an ELISA plate reader
with an upper limit of 3.0 absorbance units. The cutoff in the assay,
which was calculated as the mean absorbance of the three negative
manufacturer's controls plus 0.400 absorbance unit, ranged from 0.438
to 0.455 (mean, 0.444). The positive controls supplied with the kits
gave a reading from 1.662 to 2.252 (mean, 1.956). Positive results were
confirmed in all subjects by retesting in duplicate or triplicate, as
prescribed by the manufacturer. Furthermore, the ELISA-positive sera
were reexamined by the Chiron HCV RIBA (Chiron Corp., Emeryville,
CA), distributed by Ortho Diagnostic Systems. This assay includes the
three recombinant antigens, i.e., SOD-HCV fusion polypeptides 5-1-1
(synthesized in Escherichia coli). C100-3 (synthesized in yeast) and
SOD alone (synthesized in yeast), applied to a nitrocellulose strip. High
and low levels of human IgG are also incorporated onto each strip as
positive controls. Twenty /jl of the serum from each study subject were
incubated for 4 h at room temperature with the antigens on the strip,
and then the strip was reacted with goat anti-human IgG labeled with
horseradish peroxidase. The test was done according to the manufac
turer's instructions. The results are read by comparing the colors of the
antigen bands with those of the positive controls: negative, no visible
band; ±,visible but intensity less than low level control; 1+, intensity
Table I Démographiecharacteristics of study subjects
equal to low level control; 2+, intensity greater than low level but less
than high level control; 3+, intensity equal to high level control; 4+
Males
Females
intensity greater than high level control. Results of 1+ or greater were
Characteristic
HCC LC Controls HCC LC Control*
regarded as positive for each antigen. According to the manufacturer's
No. of subjects
73
48
291
18
27
119
instructions, samples positive for both 5-1-1 and C100-3 are "reactive";
those positive for either 5-1-1 or C100-3 (or for SOD and one or both
Age groups (no.)
of the HCV antigens) are "indeterminate"; and samples positive for
40-49
7
6
52
5
15
50-59
SOD and/or none of the HCV antigens are "nonreactive."
37
27
126
14
66
60-69
29
15
113
8
38
Statistical Analysis. \- and Fisher's exact tests were used for the
unadjusted
comparison of study groups. A Mantel-Haenszel test (18)
Median age (yr)
59.0 56.0
57.0
56.5 56.0
56.0
was performed for a stratified analysis with a few factors controlled.
Occupation
(no.)
The RRs (more correctly odds ratios) of HCC or LC for selected risk
Professional/administrative
16
8
29
10
5
factors were estimated by modeling the data through unconditional
Clerical/sales
23
13
138
12
17
71
Blue-collar
34
27
124
5
7
30
logistic regression for controlling possible confounding factors such as
Housewives
0
3
13
sex. age, etc., using the SAS statistical package (19, 20). In the model,
data on either patients with HCC or LC as well as controls were
Median VTof education
11.0 11.0
11.0
8.5
9.0
10.0
included, and the RRs of HCC or LC were calculated. The natural
2843
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
HEPATITIS C AND HEPATOCELLtJLAR
antilogarithm of regression coefficients of indicator variables is inter
preted as the RR for HCC or LC, and the 95% CI was similarly
obtained from the regression coefficient and its standard error. All
reported P values are two-tailed.
RESULTS
Sixty-two (68%) of 91 patients with HCC, 48 (64%) of 75
patients with LC, and 28 (7%) of 410 controls were found to
be positive for anti-HCV in the ELISA. In our serum samples,
positive rates in the ELISA were not associated with storage
duration of the sera or with the departments (surgery or internal
medicine) to which the patients were admitted (data not shown).
Table 2 shows the distribution of the ELISA-positive subjects
according to absorbance values in the ELISA and the results
from the RIBA. Among the ELISA-positive subjects, 46 (74%)
of 62 patients with HCC, 38 (79%) of 48 patients with LC, and
12 (43%) of 28 controls were positive for both C100-3 and 51-1, i.e., "reactive" in the RIBA. Reactive rates in the RIBA
were closely correlated with absorbance values in the ELISA;
among the subjects with absorbance values less than 2.0, only
1 control was RIBA reactive, whereas, among those with 2.0 or
greater absorbance values, the reactive rates were 85, 88, and
58% for HCC, LC, and controls, respectively. No visible bands
for SOD in the RIBA were detected through our examinations.
Table 3 presents the prevalence of anti-HCV among study
subjects by results from ELISA and RIBA and by sex. The
prevalence of RIBA-reactive results among patients with HCC,
patients with LC, and the controls were 58.9, 58.3, and 3.8%,
respectively, for males and 16.7, 37.0, and 0.8%, respectively,
for females. The difference in these figures was statistically
highly significant both for males (x2 = 161.6, d.f. = 2, P <
0.001) and for females (x2 = 38.6, d.f. = 2, P < 0.001), and
that was the case when stratified by age group (data not shown).
The difference observed was due to a higher prevalence of antiHCV among patients with HCC or LC than among controls;
anti-HCV status among patients with HCC was not signifi
cantly different from that among patients with LC. The RIBAreactive rates were significantly higher in males than in females
after being adjusted for age and study group (Mantel-Haenszel
X2= 14.!,/>< 0.001).
Table 4 shows the distribution and the prevalence of RIBA
anti-HCV by selected risk factors for HCC and LC. Only RIBAreactive results were included in Table 4. Among those subjects
positive for serum HBsAg, the prevalence of anti-HCV ranged
from 0 to 20%, which was lower than that shown in Table 3.
Among subjects with positive history of blood transfusion, 8
patients with HCC and 17 patients with LC experienced blood
transfusions related to liver diseases, e.g., due to bleeding from
esophageal varices, etc., within the past decade or so. Since
Table 2 Distribution of study subjects positive for anti-HCV in ELISA by
absorbance values and by results from RIBA
Absorbance in ELISA
Study groups
-0.9
1.0-1.9
2.0-2.9
3.0+
Total
no.
HCC
LC2;0;0;0"
0;1;0;02:3:1:01:3:0:00:1:1:11
0:1:0:40:2:4:35
0:1:3:3462
48
Controls
2:4:1:1
0:2:2:1
0:3:1:10
28
1:0:0:0
" Number of study subjects by results from RIBA in the following order:
nonreactive; 5-l-l(+) alone: C100-3(+) alone: both positive (reactive). Of the 21
subjects with 5-1-H+) alone. 18 subjects showed C100-3(±)results in RIBA, and
2 patients with HCC (absorbance. 1.30. 1.07) and I patient with LC (absorbance.
l.28)showedCIOO-3(-).
CARCINOMA
their histories of blood transfusion appeared not to be causative
events but to be the results of preexisting liver diseases such as
cirrhosis, they were not included in the blood recipients here
after. Among male blood recipients, about 70% of patients with
HCC or LC were anti-HCV reactive in contrast to 13% of
controls (P < 0.01). Among females, such an association was
less clear, but the number of subjects was limited. One notable
finding was the high prevalence (46% to 75%) of anti-HCV
among male patients with HCC or LC who had been heavy
drinkers (80 ml or more of ethanol per day for at least 10 years)
or who had no identifiable risk factors. Such values were com
parable to those detected in blood recipients with HCC or LC.
The anti-HCV status was significantly different among the
study groups with positive history of heavy drinking or with no
identifiable risk factors. Those differences arose from the higher
prevalence of anti-HCV among patients with HCC or LC than
controls; no significant difference of anti-HCV status was ob
served, however, between the HCC and LC groups.
Table 5 displays the distribution of study subjects by HBsAg
status and RIBA anti-HCV status. Among patients with HCC
or LC, there was a highly negative correlation between HBsAg
status and anti-HCV status (P < 0.05); patients with HCC or
LC positive for HBsAg tended to be much less reactive for
RIBA anti-HCV than those negative for HBsAg. Such a rela
tionship was not significant in the controls, but no controls
with positive HBsAg were RIBA reactive. Of the 2 patients
with HBsAg-positive and RIBA-reactive results, 1 patient with
HCC had received a transfusion.
The RRs of HCC and LC for selected risk factors were
estimated with and without adjustment for RIBA anti-HCV
status (Table 6). All RRs were adjusted for sex, age, occupation,
and years of education regardless of anti-HCV status. The RRs
(and 95% CIs) of HCC and LC for individuals with RIBA antiHCV reactive results compared to those with ELISA-negative
or RIBA-nonreactive results were 52.3 (23.9-114.3) and 64.4
(27.4-151.4), respectively. Those values exceeded the RRs for
chronic HBV infection as indicated by positive serum HBsAg
(15.3 for HCC; 6.1 for LC). It was difficult to adjust the antiHCV status in estimating the RR for chronic HBV infection
because of the highly negative correlation between them as
shown in Table 5. Among blood recipients, the RRs of HCC
and LC were 3.8 and 4.7, respectively, without adjustment for
anti-HCV. After an adjustment was made for anti-HCV, those
figures decreased to 2.0 and 2.7, respectively. There was a
moderate to slight risk increase among heavy drinkers (RR for
HCC, 2.2; RR for LC, 1.4). This association did not change
significantly after being controlled for anti-HCV.
DISCUSSION
This controlled study revealed an extremely strong relation
ship between anti-HCV status and HCC or LC in the Japanese
population. The RR of HCC or LC for the presence of antiHCV exceeded that for positive HBsAg. Yu et al. (21) reported
that the RR of HCC for positive anti-HCV in the ELISA was
10.5, but our estimate (52.3) was considerably higher than their
figure. The same strong association has been demonstrated in
Spain (22), Italy (23), and South Africa (24).
Johnson and Williams (25) recently expressed doubts as to
the reliability of anti-HCV results in the ELISA among patients
with HCC. They suspected that the higher prevalence of ELISA
anti-HCV among those patients could only be false positive
results due to hyperglobulinemia which is often seen in HCC
2844
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
HEPATITIS C AND HEPATOCELLULAR CARCINOMA
inELISA Anti-HCV
Table 3 Prevalence ofanli-HCV among study subjects by results from ELISA and RIBA and by sex
l'i)LC15(31.3)
of males
(%)LC12(44.4)
of females
RIBANot
+
+
+
tested
Nonreactive
Indeterminate
ReactiveTotalHCC20(27.4)
2 (2.7)
(11.0)43
g
(58.9)73(100)No.
(92.4)
1(2.1)4
2 (0.7)
(8.3)
9(3.1)
(58.3)48(100)C'ontrols269
28
(3.8)291
11
(50.0)
2(11.1)4(22.2)
0 (0.0)
1 (0.8)
5(18.5)
4 (3.4)
3(16.7)18(100)No. 10(37.0)27(100)Controls113(95.0)
(0.8)119(100)
1
(100)HCC9
Table 4 Prevalence oj RIBA anli-HCI among study subjects by selected risk factors for HCC and LC
(d.f.
2)°1.716.0'
=
FactorHBsAg-posilivePositive
(7.10
FemalesMales
(0.0)13/19
0/5
(20.0)
0/3
(0.0)8/12(66.7)
(0.0)
0/1
(0.0)3/23
FemalesMales
(68.4)
(0.0)16/26
0/3
(33.3)6/13(46.2)
3/9
(13.0)
(8.3)1/59
1/12
FemalesMales
(61.5)
(0.0)18/24
0/1
transfusion1'Positive
history of blood
drinking^Without
history of heavy
any factors aboveSexMales
(75.0)
FemalesHCC1/14* 3/10 (30.0)LC1/5
" x2 for independence of anti-HCV status among study groups.
* Number of RIBA anti-HCV reactives/number of subjects tested.
' Prevalence (r<) of RIBA ami-HC'V in parentheses.
**Histories of blood transfusion due to liver diseases «ereexcluded (see text).
(0.0)17/26(65.4)
0/1
3.040.3'131.6'
(1.7)
0/4
(0.0)7/207
(3.4)
7/15(46.7)Controls0/7 0/102(0.0)X2
46.6'
..
f Heavy drinking was defined as having consumed 80 ml or more of ethanol per day for at least 10 years.
Table 5 Distribution of study subjects by HBsAg status and RIBA anti-HCV
status"
individuals with no identifiable risk factors as well as blood
recipients among male patients with HCC or LC. Although
HCC
LC
Controls
HCV was found as a transfusion-related agent at first, there
HBsAg
anti-HCV
No.
No.
No.
%
may be other routes of transmission. Furthermore, HCV may
+
1+
+*
play a role in patients with LC which is presumably due to
18+
-f
heavy drinking and patients with HCC after suffering from so45271.119.849.529.71737301.39.349.340.008123900.02.02.995.1
called "alcoholic" cirrhosis. There was no significant difference
in anti-HCV status between the HCC and LC groups.
91
100
75
100
410
100
Total
A highly negative correlation between anti-HCV status and
r*
<().OOOI
0.028
1.0
HBsAg status was observed among patients with HCC or LC
°Both sexes were combined.
in the present study; the prevalence of anti-HCV was much
'RIBA reactive.
lower among HBsAg-positive patients with HCC or LC than
c ELISA negative or RIBA nonreactive or indeterminate.
d P for independence of HBsAg status and anti-HCV status by Fisher's exact
among HBsAg-negative ones. Among the controls such a rela
test (two tailed).
tion was not clear, but the prevalence of HBsAg and anti-HCV
in the control group might ha' c been too low to detect such an
association.
Evidence for a negative correlation between antipatients who have coexistent chronic liver diseases. Therefore,
we examined the correlation between absorbance values in the HCV status and HBsAg status has not been found in reports
ELISA and serum globulin concentration among patients with from other countries (22-24), but in Japan similar results have
been obtained (28-31). The reason of this discrepancy cannot
HCC (Fig. 1). Only a weak correlation was found (Kendall rank
be explained at present, and further investigations are needed.
correlation coefficient, 0.23; P = 0.006), and it is unlikely,
Yu et al. (21) found a synergistic effect of past or current HBV
according to our data, that the higher prevalence of anti-HCV
infection and HCV infection on HCC risk, but we could not
is simply due to hyperglobulinemia. Similar results were ob
evaluate such an effect.
tained for patients with LC (Kendall rank correlation coeffi
Histories of blood transfusion and heavy drinking were also
cient, 0.24; P = 0.006). Also, we retested the ELISA-positive
sera by the RIBA, in which reactive results have been shown to risk factors for HCC and LC in this study. After an adjustment
was made for anti-HCV, the RR for blood recipients was
be closely associated with the infectivity of HCV (26) or detec
tion of HCV RNA by polymerase chain reaction (27). In the reduced substantially but remained elevated (about 2-3-fold as
present study, more than 70% of the ELISA-positive sera from compared with nonrecipients). Thus, the assays available at
present may be undetectable for part of HCV infection, or there
patients with HCC or LC were RIBA reactive. These results
highly support the significant involvement of HCV infection in may be other transfusion-related agents. By contrast, the RR
for positive history of heavy drinking changed little after being
the etiology of HCC and LC in the Japanese population.
In this study, anti-HCV was more prevalent in males than in controlled for anti-HCV. Taking the overall high prevalence of
females. Although this phenomenon has not been observed in anti-HCV among patients with HCC or LC into consideration,
other studies in foreign countries (21-24), the same tendency
heavy alcohol consumption may promote hepatocarcinogenesis
of anti-HCV with respect to sex has been found among blood
among those infected with HCV or other viruses.
Table 7 shows the prevalence of serum anti-HCV in the
donors in Fukuoka prefecture (see later). Also noted was the
high prevalence (46-75%) of anti-HCV in heavy drinkers and
ELISA among blood donors in Fukuoka prefecture. The prev2845
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
HEPATITIS C AND HEPATOCELLULAR CARCINOMA
Table 6 Relative risks (with 95% confluence intervals) ofHCC and LC for selected risk factors with and without adjustment for RIBA anti-HCV
(95%Anti-HCVnot
FactorAnti-HCVIndeterminate
adjusted11.8(4.8-28.9)
adjusted10.4(3.9-28.1)
ReactiveHBsAg
52.3(23.9-114.3)15.3(6.1-38.1)3.8
64.4(27.4-151.4)6.1
(2.2-17.2)4.7
positivePositive
(0.9-4.7)2.1
(2.0-7.0)2.2(1.3-3.8)CDofHCCAnti-HCVadjusted:*2.0
transfusionPositive
history of blood
(2.5-8.8)1.4(0.7-2.8)CI)ofLCAnti-HCVadjusted:*2.7(1.2-6.2)1.8(0.7-4.3)
(1.0-4.4)RR°(95%Anti-HCVnot
history of heavy drinkingRR°
" Adjusted for sex. age, occupation, and years of education. For anti-HCV. individuals with ELISA-negative or RIBA-nonreactive results as the referent. For the
other factors, those negative for the corresponding factor as the referent.
* RR estimate was extremely biased due to the highly negative correlation between HBsAg status and anti-HCV status as shown in Table 5.
23.0
uat
(36). When only RIBA-reactive results were regarded as signif
icant in HCV infection, the PARs (and 95% CIs) of HCC and
LC for HCV infection were estimated as 0.49 (0.39-0.60) and
0.50 (0.38-0.61), respectively, whereas the corresponding val
ues for chronic HBV infection as indicated by positive HBsAg
were 0.20 (0.11 -0.28) and 0.09 (0.02-0.16), respectively. When
RIBA-indeterminate as well as -reactive results were considered
as meaningful, the PARs of HCV infection were calculated as
0.62 (0.51-0.72) for HCC and 0.61 (0.49-0.72) for LC. This
means that more than one-half of HCC or LC could be attrib
uted to HCV infection. Thus, the possible role of HCV in the
etiology of HCC and LC in Japan is extremely large and
possibly more important than chronic HBV infection.
lit
2.0
1.0
0.5
£?0.2
0.1
0.05-
ACKNOWLEDGMENTS
0.01
3.0
4.0
5.0
Globulin (g/dl)
We are grateful to T. Hayashi, N. Hashimoto, and M. Matsushige
for interviewing subjects; to I. Hirohata of the Department of Public
Health, Kurume University School of Medicine for preparing question
naires and advice on our interviews; and to B. T. Quinn from Kyushu
University for proofreading the manuscript. We also express our deep
appreciation to Drs. S. Sakamoto, B. Ando, and H. Miyata of the Third
Department of Internal Medicine; Dr. S. Kitano of the Second Depart
ment of Surgery, Kyushu University School of Medicine; and the staff
of all the relevant departments for their kind cooperation.
6.0
Fig. 1. Log plot of absorbance values in ELISA versus serum globulin concen
tration among patients with HCC.
, mean cutoff (0.444) in the assay.
Table 7 Prevalence of serum anti-HCt ' in ELISA among blood donors in
Fukuoka prefecture from November 1989 to March 1990
MalesAgegroups
(yr)16-19
20-29
30-39
40-49
50-59
60-64No.
All ages
of
subjects7,668
HCV(+)No.27
HCV(-I-)No.34
7.614
41
9,434
96
1.0
5,002
66
147
1.52.2
9,688
5,400
146
7,221
158
4,556
156
4.7
3.938
184
1,240Anti- 43%0.4
1.195Anti- 54%0.4 4.5FemalesNo.
39,144
666
1.7
REFERENCES
of
subjects8.027
31,839
486
1. Arthur. M. J. P., Hall. A. J.. and Wright. R. Hepatitis B, hcpatocellular
carcinoma, and strategies for prevention. Lancet, /: 607-610. 1984.
0.5
1.3
2.7
3.4
3.5
1.5
alence of anti-HCV increases with age, and it is higher in males
than in females especially among the population in their 50s or
older. The findings regarding sex correspond to our results.
The prevalence of anti-HCV among elderly blood donors was
considerably high, e.g., 4.7% for male donors ages 50 to 59
years, in comparison with the previous reports in other coun
tries (32-35). HCV infection may be responsible for high HCC
risk in this area.
We tentatively calculated the PAR (or etiological fraction,
attributable fraction), i.e., the proportion of patients with a
particular disease that can be attributed to a specified etiological
factor, for HCV infection as well as chronic HBV infection.
The derivation of PAR has been reported in detail by Greenland
2. Blumberg. B. S., and London. W. T. Hepatitis B virus and the prevention of
primary cancer of the liver. J. Nail. Cancer Inst.. 74: 267-273, 1985.
3. Beasley, R. P. Hepatitis B virus: the major etiology of hepatocellular carci
noma. Cancer (Phila.), 61: 1942-1956. 1988.
4. Szmuness, W. Hepatocellular carcinoma and the hepatitis B virus: evidence
for a causal association. Prog. Med. Virol., 24: 40-69, 1978.
5. Beasley, R. P., Hwang, L-Y.. Lin, C-C.. and Chien, C-S. Hepatocellular
carcinoma and hepatitis B virus: a prospective study of 22,707 men in Taiwan.
Lancet, 2: 1129-1133. 1981.
6. Muir. C., Waterhouse, J., Mack, T.. Powell, J., and Whelan, S. (eds.). Cancer
Incidence in Five Continents, Vol. 5. Lyon, France: International Agency for
Research on Cancer, 1987.
7. Liver Cancer Study Group of Japan. Survey and follow-up study of primary
liver cancer in Japan—Report 8 (in Japanese). Acta Hepatol. Jpn.,29: 16191626. 1988.
8. Brechot. C.. Degos. F.. Lugassy. C., Thiers. V., Zafrani. S., Franco, D..
Bismuth. H., Tròpo.C.. Bcnhamou. J-P., Wands, J.. Isselbacher. K., Tiollais,
P., and Berthelot, P. Hepatitis B virus DNA in patients with chronic liver
disease and negative tests for hepatitis B surface antigen. N. Engl. J. Med.,
312: 270-276, 1985.
9. Hiño,O., Kitagawa, T., and Sugano, H. Relationship between serum and
histochemical markers for hepatitis B virus and rate of viral integration in
hepatocellular carcinomas in Japan. Int. J. Cancer. 35: 5-10, 1985.
10. Sakamoto, M.. Hirohashi. S.. Tsuda, H., Ino. Y.. Shimosato. Y.. Yamasaki,
S., Makuuchi, M., Hasegawa, H., Terada, M., and Hosoda, Y. Increasing
incidence of hepatocellular carcinoma possibly associated with non-A, nonB hepatitis in Japan, disclosed by hepatitis B virus DNA analysis of surgically
2846
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
HEPATITIS C AND HEPATOCELLULAR
resected cases. Cancer Res.. 48: 7294-7297, 1988.
11. Okuda. K.. Fujimoto. I.. Hanai. A., and Urano. Y. Changing incidence of
hepalocellular carcinoma in Japan. Cancer Res.. 47: 4967-4972. 1987.
12. l.in.ik.i. K., Hirohata. T.. and Takeshita. S. Blood transfusion, alcohol
consumption, and cigarette smoking in causation of hepatocellular carci
noma: a case-control study in Fukuoka. Japan. Jpn. J. Cancer Res.. 79:
1075-1082. 1988.
13. Choo. Q-L.. Kuo. G.. Weiner, A. J.. Ovcrby. L. R.. Bradley. D. \V.. and
Houghton. M. Isolation of a cDNA elone derived from a blood-borne nonA. non-B viral hepatitis genome. Science (Washington DC). 244: 359-362.
1989.
14. Kuo. G.. Choo. Q-L.. Alter. H. J.. Gitnick. G. L.. Redeker. A. G.. Purcell.
R. H.. Miyamura. T.. Dienstag. J. L...Alter. M. J.. Stevens. C. E.. Tegtmeier.
G. E.. Bonino. F.. Colombo. M., Lee. \V-S.. Kuo. C'.. Berger. K.. Shusler, J.
15.
16.
17.
18.
19.
20.
R.. Overby. L. R.. Bradley. D. W.. and Houghton. M. An assay for circulating
antibodies to a major etiologic virus of human non-A. non-B hepatitis.
Science (Washington DC'). 244: 362-364. 1989.
Statistics and Information Department. Minister's Secretariat. Ministry of
Health and Welfare. Mortality Statistics from Malignant Neoplasms 19721984. Special Report of Vital Statistics in Japan (in Japanese). Tokyo:
Ministry of Health and Welfare. 1986.
Okuda. K.. Obata. H.. Jinnouchi. S.. Kubo. V.. Nagasaki. V.. Shimokawa.
Y.. Motoike. Y., Muto. H., Nakajima. Y.. Musha, H.. Yamazaki. T.,
Sakamoto. K.. Kojiro. M.. and Nakashima. T. Angiographie assessment of
gross anatomy of hepatocellular carcinoma: comparison of celiac angiograms
and liver pathology in 100 cases. Radiology. 123: 21-29, 1977.
Di Lelio, A., Cestari, C.. Loma/./.i. A., and Beretta. L. Cirrhosis: diagnosis
with sonographic study of the liver surface. Radiology. 172: 389-392. 1989.
Mantel. N.. and Haenszel. W. Statistical aspects of the analysis of data from
retrospective studies of disease. J. Nati. Cancer Inst.. 22: 719-748. 1959.
Breslow, N. E.. and Day. N. E. Statistical Methods in Cancer Research, Vol.
1. Lyon. France: International Agency for Research on Cancer. 1980.
Harrell. F. E.. Jr. The LOGIST procedure. In: S. P. Joyner (ed.). SUGI
Supplemental Library User's Guide, pp. 181-202. Cary, NC: Statistical
Analysis System Institute. Inc.. 1983.
21. Yu. M. C., Tong. M. J.. Coursaget. P.. Ross. R. K.. Govindarajan. S.. and
Henderson. B. E. Prevalence of hepatitis B and C viral markers in black and
white patients with hepatocellular carcinoma in the United States. J. Nati.
Cancer Inst.. 82: 1038-1041. 1990.
22. Bruix, J.. Barrera. J. M.. Calvet, X.. Ercilla. G.. Costa. J.. Sanchez-Tapias,
J. M.. Ventura. M.. Vail. M.. Bruguera. M.. Bru. C.. Castillo. R.. and Rodes,
J. Prevalence of antibodies to hepatitis C virus in Spanish patients with
hepatocellular carcinoma and hepatic cirrhosis. Lancet. 2: 1004-1006. 1989.
23. Colombo, M.. Kuo. G.. Choo. Q-L., Donato, M. F., Del Ninno. E.,
CARCINOMA
Tommasini. M. A.. Dioguardi. N., and Houghton. M. Prevalence of antibod
ies to hepatitis C virus in Italian patients with hepatocellular carcinoma.
Lancet, 2: 1006-1008. 1989.
24. Kew, M. C.. Houghton. M.. Choo, Q-L.. and Kuo. G. Hepatitis C virus
antibodies in southern African blacks with hepatocellular carcinoma. Lancet.
/: 873-874. 1990.
Johnson.
P. J.. and Williams. R. Hepatitis C' antibodies and hcpatocellular
25.
carcinoma: new clues or a false trail? J. Nati. Cancer Inst.. 82: 986-987.
1990.
26. Ebeling, F.. Naukkarinen. R.. and Leikola. J. Recombinant imnuinoblot
assay for hepatitis C virus antibody as predictor of inl'cctmty. Lancet. /:
982-983. 1990.
27. Weiner. A. J.. Truett. M. A.. Rosenblatt. J.. Han. J.. Quan. S.. Polito. A. J..
Kuo. G.. Choo. Q-t... Houghton. M.. Agius. C.. Page. F... and Nelles. M. J.
HCV testing in low-risk population. Lancet. 2: 695. 1990.
Akahane,
Y.. Naito. S.. Miyaz.aki. Y'.. Ainota. T.. and Suzuki. 11. Detection
28.
of anti-hepatitis C virus antibody (anti-HC'V). using C100-3 antigen (Chiron
Corp.). in the sera of non-A, non-B hepatitis (in Japanese). J. Med. Phann.
Sci., 22:601-605. 1989.
29. Kiyosawa. K.. Tanaka. E.. Sodeyama. T., Gibo. Y.. Nakano. Y.. Hayada. T..
Furuta. K.. Llsuda. S.. Shimizu. S.. Seki. K.. Nakalsuji. Y.. Furuta, S..
Watanabe. J.. and Nishioka. K. Detection of anti-hepatitis C' virus antibody
(HCV antibody) using "ORTHO HCV antibody ELISA test system" in
patients with various liver diseases and its clinical significance (in Japanese).
J. Med. Pharm. Sci.. 22: 607-612. 1989.
30. Yano, M.. Koga. M.. Inoue. C.. and Yatsuhashi. H. Clinical evaluation of
anti-hepatitis C virus antibody (HCV antibody) (in Japanese). J. Med. Pharm.
Sci.. 22:613-619. 1989.
31. Ohkoshi. S., Kojima. H.. Tawaraya. IL. Miyajinia. T.. Tomoteru. K..
Asakura, H.. Satoh. A.. Hirose. S.. Hijikata. M., Nobuyufci. K.. and Shimotohno. K. Prevalence of antibody against non-A. non-B hepatitis virus in
Japanese patients with hepatocellular carcinoma. Jpn. J. Cancer Res., 81:
550-553. 1990.
32. Kuhnl, P.. Seidl. S.. Stangel. W'.. Beyer, J.. Sibrowski. W.. and Flik. J.
Antibody to hepatitis C virus in German blood donors. Lancet. 2: 324. 1989.
33. Janot. C.. Couroucé.A. M., and Maniez. M. Antibodies to hepatitis C virus
in French blood donors. Lancet. 2: 796-797, 1989.
34. Sirchia. G., Bellobuono. A., Giovanetti. A., and Marconi. M. Antibodies to
hepatitis C virus in Italian blood donors. Lancet, 2: 797. 1989.
35. Stevens, C. E., Taylor, P. E.. Pindyck. J.. Choo. Q-L.. Bradley. D. W.. Kuo.
G.. and Houghton. M. Epidemiology of hepatitis C virus: a preliminary study
in volunteer blood donors. JAMA. 263: 49-53. 1990.
36. Greenland. S. Variance estimators for attributable fraction estimates consist
ent in both large strata and sparse data. Slat. Med.. 6: 701-708. 1987.
2847
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.
Hepatitis C and Hepatitis B in the Etiology of Hepatocellular
Carcinoma in the Japanese Population
Keitaro Tanaka, Tomio Hirohata, Shunichi Koga, et al.
Cancer Res 1991;51:2842-2847.
Updated version
E-mail alerts
Reprints and
Subscriptions
Permissions
Access the most recent version of this article at:
http://cancerres.aacrjournals.org/content/51/11/2842
Sign up to receive free email-alerts related to this article or journal.
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Department at [email protected].
To request permission to re-use all or part of this article, contact the AACR Publications
Department at [email protected].
Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1991 American Association for Cancer Research.