Gt\S’I’KOENTEKOLO(;~ 1985;88:768-72

Serum Levels of Estrogens and

Testosterone in Cirrhotic Men With and
Without Hepatocellular Carcinoma

NAOFUMI NAGASUE, YUICHIRO OGAWA, HIROFUMI YUKAYA,

NOBUHIRO OHTA, and AK10 IT0
Departments of Surgery and Internal Medicine, Hiroshima Red Cross Hospital and Atomic. Bomb
Hospital, Hiroshima, Japan and Mitsubishi Yuka Laboratory of Medical Science, Itabashi. ‘Tokyo.
Japan

Serum levels of estrogens and testosterone were liver interaction. During recent years, several work-
measured in 25 male patients with hepatocellular ers have demonstrated that estrogen receptors exist
carcinoma and associated cirrhosis of the liver and in mammalian livers (1,~). More recently, it has been
in another 25 male patients with cirrhosis only. The shown that human livers also contain estrogen re-
two groups were statistically comparable in terms of ceptors (3,4).
age distribution, duration of liver disease, incidence At present, there is evidence that oral contracep-
of alcohol abuse, incidence of hepatitis B surface tives, which usually contain estrogen and progestin.
antigenemia, and grade of hepatic dysfunction. Es- induce benign (5-8) and probably malignant (g-11)
trone was significantly elevated in both groups of tumors of the liver. Recent experimental studies also
patients. E&radio1 concentrations were above nor- support the notion that exogenous estrogen plays a
mal in 10 patients with hepatocellular carcinoma role in development of both benign (12) and malig-
and in 11 with cirrhosis only. AJI patients had nant (13-15) hepatic tumors. Although several au-
normal concentrations of estriol. There were no thors suggest that sex-steroid hormones work as
statistical differences between the fwo groups in promoters (12,13,15), a study group in Japan has
either individual or total estrogen levels (estrone demonstrated that these hormones may be initiators
0.05 < p < 0.1). Eight of the patients with hepatocel- of hepatocarcinogenesis in rats (14).
lular carcinoma and 5 of the cirrhotics had lower Most hepatocellular carcinoma (HCC) in Japan
testosterone levels than normal, but this difference occurs in men and is associated with underlying
was not significant. However, the estrone to testos- cirrhosis of the liver. Male patients with alcoholic
terone ratios were significantly higher in the hepato- cirrhosis are known to have elevated levels of estro-
cellular carcinoma group than in the cirrhosis group gens in the blood (16-181, while men with nonalco-
(p < 0.05). The present study seems to indicate that holic cirrhosis usually do not have elevated estro-
hyperestrogenemia commonly seen in male patients gens unless they are very ill (19). Therefore, it seems
with liver cirrhosis may play some role in hepatic important to determine if cirrhotic men with persis-
carcinogenesis of cirrhotic livers. Further studies are tent hyperestrogenemia have a higher risk for devel-
needed to determine if the estrone to testosterone oping hepatic tumors.
ratio is implicated in hepatocarcinogenesis in cir- In this context, we have measured serum levels of
rhotic men. estrogens [estrone (E,), estradiol (E2), estriol (EL,)]
and testostrone in male patients with liver cirrhosis
Changes in liver function and morphology after alone and in those with cirrhosis and HCC. If estro-
estrogen administration have been noted over the gens act as promoters or initiators of HCC, a higher
past decade. It is, however, uncertain if the effects of serum estrogen level would be expected in cirrhotic
estrogens on the liver are caused by a direct estrogen- patients who developed cancer compared with con-
trols. Further, the aim of this study was to evaluate
Received July 17, 1984. Accepted September 27. 1984. whether HCC would develop more frequently in the
Address requests for reprints to: Naofumi Nagasue. M.D.. De-
partment of Surgery, Hiroshima Red Cross Hospital, Hiroshima
730, Japan. Abbreviations used in this puper: E,, estrone: E,. estradiol: E II
6 1985 by the American Gastroenterological Association estriol; HCC, hepatocellular carcinoma.
0016.5085/85/$3.30
March 1985 SEX HDRMONES IN HEPATOCELLI!l,~~R CARCINOMA 769

cirrhotic patients with higher estrogen levels than in Table 1. Clinical Comparisons of Patients With
those with lower concentrations. Hepatocellular Carcinoma and Liver Cirrhosis
and Those With Cirrhosis Alone
HCC +
cirrhosis Cirrhosis
Materials and Methods Data (n = 25) (n = 25)
From July 1982 to December 1983, serum levels of Duration of liver disease” (yr) 13.1 k 8.7 10.5 + 9.9
El, EZ. Es, and testosterone were measured in 25 consecu- Serum hepatitis B virus
tive male patients with HCC and associated cirrhosis of the Hepatitis B surface antigen
liver and in 25 male patients with cirrhosis alone. Hor- (+I 3 5
mone estimations were performed before any specific (-1 20 20
treatment was used. Hepatitis B surface antibody
(+I 6 12
Age ranged from 36 to 66 yr with an average of 53.6 yr in
C-1 19 13
the HCC patients and from 38 to 67 yr with an average of
Habitual alcohol drinking” 16 14
49.1 yr in the cirrhotic patients. There was no statistical
difference in age distribution between the two groups. HCC, hepatocellular carcinoma. ‘I Mean + SD. ‘IThose who COP
Duration of liver disease before hormone assays was 13.4 sume >5O g of alcohol every day for --5 yr
yr in the HCC group and 10.5 yr in the cirrhosis group.
Serum hepatitis B surface antigen, which was measured by
the radioimmunoassay method, was positive in 5 patients according to the Student’s t-test and the x1 analysis with a
of each group. Alcohol abuse was noted in 16 patients in single degree of freedom. A p value <0.05 was considered
the former group and in 14 in the latter group [Table 1). statistically significant.
There were no differences in these data between the two
groups.
Liver function tests in both groups are summarized in
Results
Table 2. Serum activities of aspartate aminotransferase and
alanine aminotransferase were significantly higher in the Serum concentrations of estrogens and testos-
HCC group than in the cirrhosis group. There were no terone are shown in Figure 1 and Table 3. Estrone
differences between the groups in serum levels of alkaline (E,) was significantly elevated in both groups of
phosphatase, bilirubin, albumin, and total cholesterol. No
patients, but there was no significant difference
difference was noted in indocyanine green retention rate at
between the two groups (0.05 < p < 0.1). Estradiol
15 min. As to the HCC group, 14 patients had Child’s grade
(E,) levels were above normal range in 10 patients
A, 5 Child’s grade B, and 6 Child’s grade C liver disease. In
25 patients with cirrhosis alone, 11 had Child’s A, 10 with both HCC and cirrhosis and in 11 with cirrhosis
Child’s B, and 4 Child’s C liver disease (20). There were no alone. No difference was seen between the groups.
statistical differences between the two groups. Tumor All patients had normal concentrations of serum
stage was graded by the scale described by Bengmark and estriol (EJ. Eight HCC patients and 5 cirrhotic pa-
Hafstrom (21). Sixteen patients had stage I, 6 had stage II, tients had lower testosterone levels than normal. No
and 3 had stage III disease. difference was observed between the groups. On the
contrary, there was a significant difference in the E,
Hormone Assays to testosterone ratio between the two groups (p <
0.05).
Estrogens were extracted from serum with diethyl
There were no significant correlations between El,
ether. As the cross-reactivity studies showed that none of
Ez, E,, or total estrogen and testosterone concentra-
the estrogens tested had a higher cross-reactivity than
1.8% (0.02%-1.8~0), the individual estrogens were not
tions in both the HCC and cirrhosis groups. In Figure
separated. Estradiol (E2) and testosterone in serum were 2 is shown the relationship between the hormone
measured using Estradiol Radioimmunoassay-Kit (Com- levels and grade of hepatic dysfunction. Estrone (E,)
missariat a 1’Energie Atomique, France) and Testosterone tended to increase in both groups as the grade of
Radioimmunoassay-Kit (Eiken Kagaku Co., Ltd., Japan), hepatic dysfunction advanced, however. no clear
respectively. E&one (E,) and estriol (E3) in serum were trend in E2, Er3, and testosterone levels was observed.
estimated by modified methods of Linder et al. (22). Dean No significant correlation was noted between the
et al. (23), and Erlanger et al. (24). [2,4,6,7-“H]Estrone and tumor stage and hormone concentrations.
[2,4,6,7-“H]E3 were purchased from Amersham Corp., Ar-
lington Heights, Ill. Antiserum to El was obtained from
Radioimmunoassay System Laboratories, Inc., Carson Follow-up of Cirrhotic Patients
City, Nev. and antiserum to E:, was a gift from Dr. Kan-
begawa, Teikyo University School of Medicine, Tokyo,
During the follow-up period of 6-24 mo, HCC
Japan. Intraassay and interassay variation was within 10% developed in 1 of the 25 cirrhotic patients. Initial
for all four assays. serum levels of estrogens and testosterone in this
Statistical comparisons for significance were carried out patient were within normal ranges. Estrogen levels
 770 NAGASUE ET AL. GASTKOENTEKOLOGY Vol. 88. No. :j

Table 2. Liver Function Tests in Patients With

HepatoceJJuJar Carcinoma and Liver Cirrhosis
and in Those With Cirrhosis Alone
HCC +
cirrhosis Cirrhosis
Data (n = 25) (n = 25)

AST (W/L) 99 + 10" 57 k 6"

ALT (W/L) 83 t 12" 38 + 5"
ALP (KAU) 19.7 + 3.2 15.0 -+ 1.2
Total bilirubin (mgi100 ml) 1.3 + 0.2 1.2 t 0.2
Serum albumin (g/100 ml) 3.6 + 0.1 3.7 k 1.0
Total cholesterol (mgi100 ml] 139 f 8 139 2 7
KG U&,1 (%I 27.5 k 5.2 35.3 t 2.6

ALP, alkaline phosphatase (normal range 2.0-10.0 KAU); ALT.

alanine aminotransferase (normal range 3-32 W/L); AST, aspar-
tate aminotransferase (normal range O-37 W/L); HCC, hepatocel-
lular carcinoma; ICG, indocyanine green. Values represent mean
t SD. ’ p < 0.01." p < 0.05.

t
had been elevated 17 mo later when HCC was I
ESTRONE ESTRAOIOL ESTRIOL TESTOSTERONE0
recognized although hepatic functions had also dete- Figure 1. Serum concentrations of estrogens and testosterone in
riorated from Child’s grade A to grade C. There were 25 male patients with hepatocellular carcinoma (hepa-
7 patients whose total estrogen level had been >150 toma) and liver cirrhosis and in 25 men i\rith onI>
pg/ml, but HCC occurred in none of them during the cirrhosis. Shaded areas indicate normal ranges.

follow-up period of 10-24 mo.

holic cirrhosis of the liver (16-19). As far as we are
aware, there are no data concerning how persistent
Discussion hyperestrogenemia commonly seen in cirrhotic men
influences the development and progression of HCC.
It seems almost unquestionable that oral con- In the present study, serum E, was significantly
traceptives induce benign hepatic tumors such as elevated in both groups of patients; a similar trend
liver cell adenoma or focal nodular hyperplasia (5- was observed for E2. The E, levels tended to be
8) and possibly HCC (9-11). Although there may be higher in the HCC group as compared with the
chemical differences between synthetic estrogens cirrhosis group (0.05 < p < O.l), and there was a
such as ethynyl E2 that have been implicated in oral significant difference in E, to testosterone ratios
contraceptive-related hepatic lesions and endoge- between the two groups (p < 0.05). Whether this
nous estrogens, Webster et al. (25) recently reported increased El to testosterone ratio in the HCC group is
an interesting case which supported the notion that important for development of HCC remains to be
endogenous hyperestrogenemia like exogenous sex determined. As with other factors determining estro-
steroids might also predispose to the development of gen responsiveness, it is theoretically possible that
hepatic adenoma. changes in the sex steroid binding globulin or hepat-
It has been the opinion of pathologists heretofore
that neither liver cell adenoma nor focal nodular Table 3. Serum Concentrations of Estrogens and
hyperplasia is a precursor of HCC (26). There have Testosterone in Two Groups of Patients
been several reports, however, that suggest the possi- HCC +
bility of malignant transformation of these tumors cirrhosis (Zirrhosis
(10,27,28). Although the relation between oral con- Data (n = 25) (I1 = 253

traceptives and HCC has not been statistically prov- Estrogens (pgiml)
en, it remains a probable association in the light of E, 101 f 10" 78 2 fi"
available clinical and experimental studies (13-15). E2 38 -+ 3 38 k :S
E? 12 c:!
Several workers assume that estrogens act as promot-
Testosterone (@ml) 6.8 2 0.8 7.3 e 0.7
ers in hepatocarcinogenesis (12,13,15), and there is El + E, + I&/testosterone 43.5 + 62.0 21.8 k 20.1
some evidence that estrogens may act as initiators E,itestosterone 35.1 ? 56.5" 12.1 + 7.1"
(14).
HCC, hepatocellular carcinoma. Values represent mean -t SD. E,,
On the other hand, estrogen-androgen imbalance estrone (normal range 5-40 pgiml); EL. estradiol [normal range
has been well known to occur in male patients with lo-40 pgiml); E,, estriol (normal range Cl0 pgiml); testosterone
chronic liver disease, especially in those with alco- (normal range 4-14 ngiml). ” p < 0.1. ‘J p < 0.05.
March 1985 SEX HORMONES IN HEPATOCRLLlJLAR C.%RCINOMA 771

References

1. Chamness GC. Costlow ME. McGuire WL. Estrogen receptor

in rat liver and its dependence on prolactin. Steroids 1975:
26:363-71.
2. Aten RF, Dickson RB. Eisenfeld AJ. Female and male green
monkey liver estrogen receptor. Biochem Pharmacol 1979;
28:2445-50.
3. Duffy MJ, Duffy GJ. Estradiol receptors in human liver. 1
Steroid Biochem 1978;9:23:3-5.
4. Porter LE, Elm MS. van Thiel DH. Dugas MC, Eagon PK.
Characterization and quantitation of human hepatic estrogen
receptor. Gastroenterologg 1983;84:704-12.
5. Baum JK. Bookstein JJ. Holtz F, Klein EW. Possible associa-
.;” tion between benign hepatomas and oral contraceptives.
0

.00
Lancet 1973:ii:926-9.
I 6. Edmondson HA, Reynolds TB, Henderson B, Benton B. Re-
e A B c A 0 c A B C A B C gression of liver cell adenomas associated with oral contra-
ceptives. Ann Intern Med 1977:86:180-2.
Figure 2. Relationship between sex hormone concentrations and 7. Steinbrecher UP, Lisbona R. Huang SS. Mishkin S. Complete
grade of hepatic dysfunctioli. Shaded or‘ras indicate regression of hepatocellular adenoma after withdrawal of oral
normal ranges. contraceptives. Dig Dis Sci 1981:26:1045-50.
8. Buhler H, Pirovino M, Akovbiantz A, et al. Regression of liver
cell adenoma: a follow-up study of three consecutive patients
ic estrogen receptor number and affinity may have after discontinuation of oral c.ontraccptive use. Gastroenterol-

ogy 1982;82:775-82.
altered the effect of apparently similar steroid levels
9. O’Sullivan JP, Rosswick RI’. Oral contraceptives and malig-
on the cirrhotic liver. Unfortunately, these parame-
nant hepatic tumors. Lancet 1976;i:l124-5.
ters were not measured in the present study. 10. Neuberger J, Portman B, Nunnerley HB. Laws JW, Davis hl.
During the follow-up of the 25 cirrhotic patients Williams R. Oral-contraceptive-associated liver tumors: oc-
for 6-24 mo, HCC occurred in 1 patient 17 mo after currence of malignancy and difficulties in diagnosis. Lancet
the hormone measurements. Initial estrogen levels in 198O:i:273-6.
11. Malt RA, Galdabini JJ, Jeppson BM’. Abnormal sex-steroid
this patient were almost within normal ranges. Hor-
milieu in young adults with hepatocellular carcinoma. World
mone levels had increased to above the normal J Surg 1983;7:247-52.
ranges 17 mo later when HCC was first recognized, 2. Wanless IR. Medline A. Role of estrogens as promoters of
but at this time hepatic functions had also deteriorat- hepatic neoplasia. Lab Invest 1982:46:313-2Ul.
ed from Child’s A to Child’s C. No HCC to date has 3. Taper HS. Effect of estradiol-17.phenyl-propionate and estra-
dial benzoate on N-nitrosomorpholine-induced liver carcino-
occurred in the cirrhotics with extremely high estro-
genesis in olrariectomized female rats. Cdncer 1978;42:462-7.
gen levels. It is, however, still uncertain if HCC tends 14. Higashi S, Tomita T, Mizumoto R, Nakakuki K. Development
to develop more frequently in those patients with of hepatoma in rats following oral administration of synthetic
very high estrogen levels, as the follow-up time in estrogen and progesterone. Gann 1980;71:576-7
these patients is short. 15. Yager JD Jr, Yager R. Oral contraceptive steroids <is promotors
of hepatocarcinogenesis in female Sprague-Dawley rats. Can-
An etiologic role for hepatitis B virus in hepatocar-
cer Res 1980;40:3680-5.
cinogenesis has been proposed by numerous investi- 16. Chopra IJ. Tulchinsky D, Greenway FL. Estrogen-androgen
gators, and most HCCs in Japan are thought to be imbalance in hepatic cirrhosis. Ann IntcJrn Mad 1973:79:198-
associated with hepatitis B virus (29,30). Therefore, 302.
17. Pentikainen PJ, Pentikainen LA, Azarnoff DL. IIujoune CA.
it could be argued that the present patient popula-
Plasma levels and excretion of oestrogens in urine in chronic:
tion may not be appropriate for examining the role of
liver disease. Gastroenterologq 1975:69:2&7.
estrogens in hepatocarcinogenesis inasmuch as the 18. Green JRB. Mowat NAG, Fischer KA. Anderso~~ DC. Plasma
virus may be much more potent as an inducer than oestrogens in men with chronic liver disease. Gut 1976;
are the estrogens. As shown in Table 1, however, the 17:426-30.
rate of hepatitis B surface antigenemia was similar in 19. Van Thiel DH, Gavaler JS, Spero JA. et al. Patterns of hypotha-
lamic-pituitary-gonadal dysfunction in men with liver dis-
both groups. Also, the duration of chronic liver
ease due to differing etiologies. Hepatology 1981:1:39-46.
disease before the present study was not different 20. Child CG, Turcotte JG. Surgery in portal hypertension. In:
between the two groups. Therefore, our results sug- Child CG, ed. Major problems in clinical surgery: the liver
gest the possibility that a sustained rise in serum and portal hypertension. Philadelphia: WB Saunders, 1964:
1-85.
estrogen levels, especially El, and an increase in the
21 Bengmark S, Hafstrom LO. The natural history of primary and
E, to testosterone ratio may be associated with an secondary malignant tumors of the liver. I. The prognosis for
increased risk of HCC in cirrhotic male patients. patients with hepatic metastases from colonic and rectal
Further studies on this subject are necessary to reach carcinoma verified by laparotomy. Cancer 1969:23:198-202.
a firm conclusion. 22 Linder HR. Porel E. Friedlander A, Zcaitlin A. Specificity of
772 NAGASUE ET AL. GASTROENTEROLOGY Vol. 88, No. 3

antibodies to ovarian hormones in relation to the site of 27. Davis M, Portmann B, Searle M, Wright R, Williams R.
attachment of the steroid hapten to the peptide carrier. Histological evidence of carcinoma in a hepatic tumor associ-
Steroids 1972;19:357-75. ated with oral contraceptives. Br Med J 1974;4:496-8.
23. Dean PDG. Exley D, Johnson MW. Preparation of 17-oestra- 28. Tesulk H, Lawrie J. Hepatocellular adenoma. Its transition to
diol-6-(0-carbonymethyl) oxime-bovine serum albumin con- carcinoma in a user of oral contraceptives. Arch Path01 Lab
jugate. Steroids 1971;18:593-603. Med 1981;105:296-9.
24. Erlanger BF. Borek F. Bieser SM, Lieberman S. Steroid- 29. Akagi G, Furuya K, Otsuka H. Hepatitis B antigen in the liver
protein conjugates. J Biol Chem 1957;228:713-27. in hepatocellular carcinoma in Shikoku, Japan. Cancer
25. Webster MW, Van Thiel DH, Bron KM, Barnes EL. Hepatic 1982;49:678-82.
adenoma associated with portasystemic shunting in a young 30. Hino 0, Kitagawa T, Koike K. et al. Detection of hepatitis B
woman. Digestion 1979;19:328-34. virus DNA in hepatocellular carcinoma in Japan. Hepatology
26. Ishak KG, Rabin L. Benign tumors of the liver. Med Clin North 1984;4:90-5.
Am 1975;59:995-1013.