elevated serum testosterone levels and risk of hepatocellular … · genic-anabolic steroid...

6
[CANCER RESEARCH 53. 7VO-794. February 15. 1993] Elevated Serum Testosterone Levels and Risk of Hepatocellular Carcinoma1 Ming-Whei Yu and Chien-Jen Chen2 Comprehensive Cancer Center and Division of Environmental Sciences, School of Public Health. Columbia University, New York, New York 10032 ¡M-W. Y.(, anil Institute of Public Health, College of Medicine, National Taiwan University, Taipei ¡0018. Taiwan ¡C-J.CJ ABSTRACT Serum samples of 9691 male adults had been collected and frozen for a prospective study of hepatocellular carcinoma in Taiwan. After an average follow-up period of 4.6 years, testosterone levels in the stored serum were measured by radioimmunoassay using commercial kits for 35 cases of newly developed hepatocellular carcinoma, 63 hepatitis B surface antigen- negative and 77 hepatitis B surface antigen-positive matched controls. Elevated testosterone levels were found to be associated with an increased risk of hepatocellular carcinoma. The association remained significant after the adjustment for effects of other hepatocellular carcinoma risk factors, including hepatitis B surface antigen carrier status, positivity of serum antibody to hepatitis C virus, cigarette smoking, alcohol drinking, past liver disease history, and dietary habits. The multivariate-adjusted relative risk of hepatocellular carcinoma for men with testosterone levels in the upper fertile was 4.1 (95% confidence interval = 1.3-13.2) com pared with those having levels in the middle or lower tertiles (P = 0.016). The results consistent with those observed in animal experiments support the hypothesis that testosterone plays a role in the etiology of human hepatocellular carcinoma. INTRODUCTION Liver cancer, largely HCC,1 is one of the most common fatal cancers in the world ( 1). Although there has been rapid progress in the understanding of the etiology of HCC during the past 2 decades, the reasons for marked male predominance in HCC incidence observed in both high- and low-risk areas (I, 2) remain to be elucidated. In Taiwan. HB„Agcarrier status has been well documented as the most important risk factor of HCC (3-5). However, HCC is 2-3 times more frequent in males than in females, despite their similarity in HBsAg carrier rate (2). Differences in life-style habits such as cigarette smok ing and alcohol drinking have been implicated in the gender discrep ancy (4). Sex hormone and striking male-to-female differences in chronic hepatitis (6) and liver cirrhosis (7) may also be important. A greater susceptibility of male animals to spontaneous and chem ical carcinogen-induced HCC has long been observed in experimental mice and rats (8-14, 15). The increased susceptibility to hepatocar- cinogenesis in males has also been observed in HBV transgenic mice (16, 17). These observations are in agreement with epidemiological data suggesting that men are much more prone to HCC than women. Although the marked sex difference in HCC incidence observed in mice and rats may result from the tumor-inhibiting effects of estrogen in female animals as well as the tumor-promoting effects of testos terone in male animals, the persuasive evidence in the animal models indicating that testosterone may enhance the development of HCC suggests that the testosterone may explain, at least in part, the greater Received 6/29/92; accepted 12/4/92. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Supported by the Department of Health. Executive Yuan. Republic of China. : To whom requests for reprints should be addressed, at Institute of Public Health. National Taiwan University College of Medicine. No. 1 Jen-Ai Road, Section 1. Taipei I(K)I8, Taiwan. 1The abbreviations used are: HCC. hepatocellular carcinoma; HBsAg. hepatitis B surface antigen; HBV. hepatitis B virus; HCV. hepatitis C virus; RIA. radioimmunoassay; CI. confidence interval; RPHA, reverse passive hemagglutination assay; RR. relative risk. incidence of liver tumor observed in men. Male castration signifi cantly decreases the frequency of chemically induced HCC (10, 12, 15, 18, 19), whereas administration of testosterone to castrated male rats or to female mice increases the growth of preneoplastic hepatic- foci ( 11) or the incidence of spontaneous (8) or chemical carcinogen- induced HCC ( 18). An experiment on rats also showed that hypophy- sectomy in males inhibits liver tumor induction by aflatoxin (20). In contrast to the abundant data in animal models, the evidence showing the effect of testosterone on human hepatocarcinogenesis is scanty in spite of several documented HCC cases following andro- genic-anabolic steroid treatment (21) and the presence of androgen receptors in human HCC tissues (22, 23). We therefore carried out a nested case-control study using serum samples collected and frozen in 1984-1986 from a cohort of 9691 male adults to investigate the role of elevated testosterone in the etiology of HCC. SUBJECTS AND METHODS The study population consisted of 9691 men aged 30-85 years who attended a community-based study carried out in six townships of Taiwan for the early detection of cancer between September 1984 and February 1986. At the initial recruitment examination, a total of 15 ml blood were collected from each study subject. They were also personally interviewed according to a structured ques tionnaire on sociodemographic characteristics, long-term habits of cigarette smoking, alcohol drinking, and dietary pattern, as well as personal and family history of chronic diseases by well-trained public health nurses in local health centers. Habit of cigarette smoking was defined as having smoked cigarettes more than 4 days a week for at least 6 months. Alcohol drinking was defined as having drunk alcohol more than 3 days a week for at least 6 months. Vegetable consumption frequency was defined as the number of meals includ ing fresh vegetables per week. Vegetarian habit was defined as having one or more meals without eating food from animal sources every day for more than 1 year. The personal and family history of chronic liver diseases diagnosed by physicians was also obtained. The participants of this study cohort were followed by telephone interviews and home visits annually until March 1990. The person-year under observation for each subject was defined as the period of follow-up from recruitment to the dale when he was affected with liver cancer or died from other causes or the date this nested case-control study started, i.e.. March 1990. The follow-up period ranged from 0.5 to 5.5 years, with an average of 4.6 years. Death certificates were also reviewed to examine the causes of death for those who died. During the follow-up period, 36 newly diagnosed liver cancer cases were identified. The annual incidence of liver cancer was 91 Al \OO.(KX)person- years. Although the coding of liver cancer in death certificates combined cancers of the liver and the intrahepatic bile duct (1CD 155) according to the 8th Revision of the International Classification of Diseases, Injuries and Causes of Death, more than 95% of the deaths classified into this category in Taiwan were HCC. The most accepted diagnostic criteria for HCC in Taiwan are assigned on the basis of either pathological examinations or elevated a-fetoprotein level (>40() ng/ml) combined with at least one positive image on angiography. sonography. liver scan, and/or computerized tomography scans. Among the 36 liver cancer cases, 8 were double-checked with hospital records (4 were confirmed histologically and 4 were diagnosed by an elevated a-fe toprotein level and liver images compatible to HCC). and 15 were double- checked with data files of the National Cancer Registry in Taiwan (about one-half of them were diagnosed pathologically and another half by elevated a-fetoprotein and liver images). The remaining 13 cases were identified through the national death certification system only. Since HCC is a common 790 Research. on January 7, 2021. © 1993 American Association for Cancer cancerres.aacrjournals.org Downloaded from

Upload: others

Post on 17-Sep-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Elevated Serum Testosterone Levels and Risk of Hepatocellular … · genic-anabolic steroid treatment (21) and the presence of androgen receptors in human HCC tissues (22, 23). We

[CANCER RESEARCH 53. 7VO-794. February 15. 1993]

Elevated Serum Testosterone Levels and Risk of Hepatocellular Carcinoma1

Ming-Whei Yu and Chien-Jen Chen2

Comprehensive Cancer Center and Division of Environmental Sciences, School of Public Health. Columbia University, New York, New York 10032 ¡M-W. Y.(, anil Institute ofPublic Health, College of Medicine, National Taiwan University, Taipei ¡0018.Taiwan ¡C-J.CJ

ABSTRACT

Serum samples of 9691 male adults had been collected and frozen for aprospective study of hepatocellular carcinoma in Taiwan. After an averagefollow-up period of 4.6 years, testosterone levels in the stored serum were

measured by radioimmunoassay using commercial kits for 35 cases ofnewly developed hepatocellular carcinoma, 63 hepatitis B surface antigen-negative and 77 hepatitis B surface antigen-positive matched controls.

Elevated testosterone levels were found to be associated with an increasedrisk of hepatocellular carcinoma. The association remained significantafter the adjustment for effects of other hepatocellular carcinoma riskfactors, including hepatitis B surface antigen carrier status, positivity ofserum antibody to hepatitis C virus, cigarette smoking, alcohol drinking,past liver disease history, and dietary habits. The multivariate-adjusted

relative risk of hepatocellular carcinoma for men with testosterone levelsin the upper fertile was 4.1 (95% confidence interval = 1.3-13.2) compared with those having levels in the middle or lower tertiles (P = 0.016).

The results consistent with those observed in animal experiments supportthe hypothesis that testosterone plays a role in the etiology of humanhepatocellular carcinoma.

INTRODUCTION

Liver cancer, largely HCC,1 is one of the most common fatal

cancers in the world ( 1). Although there has been rapid progress in theunderstanding of the etiology of HCC during the past 2 decades, thereasons for marked male predominance in HCC incidence observed inboth high- and low-risk areas (I, 2) remain to be elucidated. In

Taiwan. HB„Agcarrier status has been well documented as the mostimportant risk factor of HCC (3-5). However, HCC is 2-3 times more

frequent in males than in females, despite their similarity in HBsAgcarrier rate (2). Differences in life-style habits such as cigarette smok

ing and alcohol drinking have been implicated in the gender discrepancy (4). Sex hormone and striking male-to-female differences in

chronic hepatitis (6) and liver cirrhosis (7) may also be important.A greater susceptibility of male animals to spontaneous and chem

ical carcinogen-induced HCC has long been observed in experimentalmice and rats (8-14, 15). The increased susceptibility to hepatocar-

cinogenesis in males has also been observed in HBV transgenic mice(16, 17). These observations are in agreement with epidemiologicaldata suggesting that men are much more prone to HCC than women.Although the marked sex difference in HCC incidence observed inmice and rats may result from the tumor-inhibiting effects of estrogenin female animals as well as the tumor-promoting effects of testos

terone in male animals, the persuasive evidence in the animal modelsindicating that testosterone may enhance the development of HCCsuggests that the testosterone may explain, at least in part, the greater

Received 6/29/92; accepted 12/4/92.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1Supported by the Department of Health. Executive Yuan. Republic of China.: To whom requests for reprints should be addressed, at Institute of Public Health.

National Taiwan University College of Medicine. No. 1 Jen-Ai Road, Section 1. Taipei

I(K)I8, Taiwan.1The abbreviations used are: HCC. hepatocellular carcinoma; HBsAg. hepatitis B

surface antigen; HBV. hepatitis B virus; HCV. hepatitis C virus; RIA. radioimmunoassay;CI. confidence interval; RPHA, reverse passive hemagglutination assay; RR. relative risk.

incidence of liver tumor observed in men. Male castration significantly decreases the frequency of chemically induced HCC (10, 12,15, 18, 19), whereas administration of testosterone to castrated malerats or to female mice increases the growth of preneoplastic hepatic-

foci ( 11) or the incidence of spontaneous (8) or chemical carcinogen-induced HCC ( 18). An experiment on rats also showed that hypophy-

sectomy in males inhibits liver tumor induction by aflatoxin (20).In contrast to the abundant data in animal models, the evidence

showing the effect of testosterone on human hepatocarcinogenesis isscanty in spite of several documented HCC cases following andro-genic-anabolic steroid treatment (21) and the presence of androgen

receptors in human HCC tissues (22, 23). We therefore carried out anested case-control study using serum samples collected and frozen in1984-1986 from a cohort of 9691 male adults to investigate the role

of elevated testosterone in the etiology of HCC.

SUBJECTS AND METHODS

The study population consisted of 9691 men aged 30-85 years who attended

a community-based study carried out in six townships of Taiwan for the early

detection of cancer between September 1984 and February 1986. At the initialrecruitment examination, a total of 15 ml blood were collected from each studysubject. They were also personally interviewed according to a structured questionnaire on sociodemographic characteristics, long-term habits of cigarette

smoking, alcohol drinking, and dietary pattern, as well as personal and familyhistory of chronic diseases by well-trained public health nurses in local health

centers. Habit of cigarette smoking was defined as having smoked cigarettesmore than 4 days a week for at least 6 months. Alcohol drinking was definedas having drunk alcohol more than 3 days a week for at least 6 months.Vegetable consumption frequency was defined as the number of meals including fresh vegetables per week. Vegetarian habit was defined as having one ormore meals without eating food from animal sources every day for more than1 year. The personal and family history of chronic liver diseases diagnosed byphysicians was also obtained.

The participants of this study cohort were followed by telephone interviewsand home visits annually until March 1990. The person-year under observationfor each subject was defined as the period of follow-up from recruitment to the

dale when he was affected with liver cancer or died from other causes or thedate this nested case-control study started, i.e.. March 1990. The follow-up

period ranged from 0.5 to 5.5 years, with an average of 4.6 years. Deathcertificates were also reviewed to examine the causes of death for those whodied. During the follow-up period, 36 newly diagnosed liver cancer cases wereidentified. The annual incidence of liver cancer was 91 Al \OO.(KX)person-

years. Although the coding of liver cancer in death certificates combinedcancers of the liver and the intrahepatic bile duct (1CD 155) according to the8th Revision of the International Classification of Diseases, Injuries andCauses of Death, more than 95% of the deaths classified into this category inTaiwan were HCC. The most accepted diagnostic criteria for HCC in Taiwanare assigned on the basis of either pathological examinations or elevateda-fetoprotein level (>40() ng/ml) combined with at least one positive image on

angiography. sonography. liver scan, and/or computerized tomography scans.Among the 36 liver cancer cases, 8 were double-checked with hospital records(4 were confirmed histologically and 4 were diagnosed by an elevated a-fetoprotein level and liver images compatible to HCC). and 15 were double-

checked with data files of the National Cancer Registry in Taiwan (aboutone-half of them were diagnosed pathologically and another half by elevateda-fetoprotein and liver images). The remaining 13 cases were identified

through the national death certification system only. Since HCC is a common

790

Research. on January 7, 2021. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

Page 2: Elevated Serum Testosterone Levels and Risk of Hepatocellular … · genic-anabolic steroid treatment (21) and the presence of androgen receptors in human HCC tissues (22, 23). We

SERUM TESTOSTERONE AND Hi;i>AT(X'i;Ll,UI.AR CARCINOMA

cancer and most HCC patients are referred to teaching hospitals for diagnosisand treatment in Taiwan, the possibility of misdiagnosis of HCC was rather

low.Because liver cirrhosis has been reported to be frequently associated with

hypogonadism (241 and most HCC patients in Taiwan may have a long-term

underlying liver cirrhosis before they are affected by HCC. it is essential to

evaluate the effect of liver cirrhosis on serum testosterone in the preclinicalphase of HCC patients. The deep-fro/.en serum samples from IS study subjectswho died from liver cirrhosis during the 5-year follow-up period were used to

assess whether liver cirrhosis may cause a significant change in serum testos

terone levels and distort the association between serum testosterone levels andHCC risk. Since all of these cirrhotic patients died within 5 years after therecruitment but only 3 reported a history of physician-diagnosed liver cirrhosis

in the baseline interview, it was believed that these patients already had a

subclinical liver cirrhosis at the time of blood collection, the same as most

HCC patients identified in this study.All subjects in the study cohort were tested for their HBsAg carrier status by

RPHA at the initial recruitment examination. Because some HB„Agcarriersmay have chronic liver diseases, it is essential to obtain a valid estimation ofserum testosterone level for HB„Ag-positive and HB„Ag-negative subjects,

respectively. Two HBsAg-positive and two HB„Ag-negativecontrols matched

with each liver cancer case were selected in order to recruit enough and equalnumbers of subjects not affected with HCC. Controls were randomly selected

from study subjects who were alive and free from liver cancer on the dates atthe diagnosis of liver cancer cases to whom they were matched. Cases andcontrols were matched with respect to age (within 5 years), date of questionnaire interview and blood collection (within 3 months), and residential town

ship. As the frozen serum sample of one liver cancer case was not enough for

the test of testosterone, the statistical analysis was based on the data of 35matched case-control sets.

Serum samples were separated on the same day as blood collection.Two aliquots were frozen in deep freezers and transported in dry ice to the

central laboratory at National Taiwan University College of Medicine. Theserum samples were kept at -30°C for further examination until this nested

case-control study was carried out. Serum samples of 35 liver cancer cases and

140 matched controls were tested in April 1991. The HB„Agstatus of livercancer cases and matched controls who were HB„Ag-negative in the initial

RPHA were retested by RIA using commercial kits (Abbott Laboratories,North Chicago, IL). Seven controls who were HB^Ag-negative on RPHA werefound to be HBsAg-positive on RIA. In other words, there were 77 HBsAg-positive and 63 HBsAg-negative controls based on RIA results. Anti-HCV was

examined in duplicate by enzyme immunoassay (Abbott Laboratories) according to the manufacturer's instructions. Positive samples at the first test were

retested. Only repeatedly positive samples were considered anti-HCV-positive.

Serum testosterone levels were measured by RIA using commercial kits(Biomerieux. Marcy l'Etoile, France). Specimens for each liver cancer case

and the matched controls were tested blindly in the same batch for all labo

ratory examinations.A age-adjusted serum testosterone value was used when we compared the

mean levels of the hormone of liver cirrhosis cases to those of controls. Sincethe distribution of the age-adjusted testosterone values showed no substantial

deviation from the normal distribution, judging from histogram and coefficients of skewness and kurtosis. a r test was applied to compare the mean levels

of serum testosterone between groups. To assess the effect of elevated serumtestosterone on the risk of HCC. the serum testosterone level was dichotomizedusing the lowest value of the upper tertile of testosterone levels in controls, i.e.,5.69 ng/ml. as the cutoff point. The relative risk of liver cancer for the highertestosterone level was estimated by the odds ratio using the lower testosteronelevel as the referent group. Conditional logistic regression analyses were usedto calculate matched crude and multivariate-adjusted odds ratios and their 95%

CI for various risk factors of HCC. Statistical significance of the risk estimateswas examined by Z test. All P values for tests of statistical significance werebased on two-tailed tests.

This study was conducted with the approval of the Department of Healthand in compliance with regulations for the protection of human researchsubjects.

RESULTS

There were 20 HBsAg-positive and 15 HBsAg-negative liver cancer

cases. The age of liver cancer cases at the initial recruitment examination ranged from 40 to 74 years. The mean age ±SD at recruitmentwas 60.1 ±8.8, 58.6 ±8.9. 58.3 ±9.8, and 59.2 ±9.2 years,respectively, for HB„Ag-negativecontrols. HBsAg-positive controls,HBsAg-negative liver cancer cases, and HBsAg-positive liver cancer

cases. The age distribution of liver cancer cases was similar to that oftheir matched controls.

The age of patients who died from liver cirrhosis ranged from 38 to66 years, with a mean ±SD of 52.5 ±9.6 years. They were youngerthan controls. Since the serum testosterone level in men has beenreported to decline gradually with age after 40 years (25), age-ad

justed testosterone values were derived for comparison. The age andtestosterone level of controls were used to build a regression line. Aresidual was first calculated by subtracting the predicted value obtained through the regression equation from the observed testosteronelevel. The residual of each study subject was then added to the meantestosterone level of controls to derive an age-adjusted value of serumtestosterone. Since the regression coefficient (b = 0.00056, P = 0.98)

of this regression line indicated that the testosterone level did notsignificantly change with age, the adjustment would not alter theobserved association between liver cirrhosis and serum testosteronelevels. The age-adjusted serum testosterone levels of liver cirrhosis

cases and controls by HBsAg carrier status were compared in Table 1.There were 12 HBsAg-positive and 3 HBsAg-negative cases of livercirrhosis. The HB^Ag-negative liver cirrhosis cases had the lowest

testosterone level among the four groups. But the difference in meanserum testosterone level was not significant between any two groups.The proportion of a testosterone level above 5.69 ng/ml was significantly higher in HBsAg-positive controls than in HBsAg-negativecontrols (P = 0.045). The proportion was slightly higher in HBsAg-positive controls than HBNAg-positive liver cirrhosis cases.

The mean ±SD of serum testosterone levels was 5.20 ±2.52ng/ml for liver cancer patients. Among the HBsAg-positive matchedcase-control sets, 55.0% of liver cancer cases and 47.5% of matched

controls had a testosterone level greater than 5.69 ng/ml, giving amatched RR of 1.5. Among the HBsAg-negative matched sets, 46.7%

of liver cancer cases and 23.3% of matched controls had an elevatedtestosterone level, giving a matched RR of 2.3. In order to increase thestatistical power of detecting a significant effect of elevated testosterone, all controls (2 HBsAg-positive and 2 HBsAg-negative) matched

with each liver cancer case were included in the analyses. Because theHB.,Ag carrier rate of the liver cancer cases (57%) was similar to thatof controls (55%), the association between serum testosterone leveland HCC observed in this study is acceptable. The proportion ofelevated testosterone level was compared between liver cancer casesand matched controls, as shown in Table 2. Eighteen (51.4%) of 35liver cancer cases and 48 (34.3%) of 140 matched controls had aserum testosterone level greater than 5.69 ng/ml. There was a significant association between elevated serum testosterone level and livercancer, with a RR of 2.3 (95% CI = 1.01-5.14) and a P value of

0.046.

Table 1 Age-adjusted serum lesinstemne levels (ng/ml) of liver cirrhosis cases

and controls h\ HR^Ag carrier status

Controls Liver cirrhosis cases

HBsAg n Mean±SD >3.W (%) « Mean ±SD >5.69 (%)

Negative 63 4.54 ±1.87Positive 77 5.22 ±3.11

25.441.6''

3 2.92 ±1.77 012 5.24 ±2.47 33.3

" The lowest value of upper tertile in the testosterone distribution of controls.h P = 0.045 compared to HBsAg-negative controls.

791

Research. on January 7, 2021. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

Page 3: Elevated Serum Testosterone Levels and Risk of Hepatocellular … · genic-anabolic steroid treatment (21) and the presence of androgen receptors in human HCC tissues (22, 23). We

SERUM TESTOSTERONE AND HEPATOCELLULAR CARCINOMA

Table 2 Association benreen elevateti serum testosterone levelsanil the risk of HCC

Serumtestosteronelevel

(ng/ml)S5.69

>5.69CasesNo.1718<%)(48.6)(51.4)ControlsNo.9248(%)(65.7)04.3)Relative

risk(95% confidenceinterval)1.02.3(1.01-5.14)

Table 3 Relative risk of factors associated with HCC

CasesVariableAnti-HCVNegativePositiveCigarette

smokingNoYesAlcohol

drinkingNoYesVegetarian

habitNoYesNo.2871124269296(%)(80.0)(20.0)(31.4)(68.6)(74.3)(25.7)(82.9)(17.1)ControlsNo.136459811211912713(%)

(Ç(97.1)(2.9)(42.1)(57.9)(86.4)(13.6)(90.7)(9.3)Relativerisk

'5% confidenceinterval)1.011.8(2.4-57.7)1.01.6(0.7-3.5)1.02.6(1.0-7.0)1.02.2

(0.7-6.7)

Vegetable consumption (meals/week)"

>6 27 (79.4) 131 (93.6) 1.0<6 7 (20.6) 9 (6.4) 4.1(1.3-13.4)

Past liver disease historyNo 29 (82.9) 134 (95.7) 1.0Yes 6 (17.1) 6 (4.3) 5.0(1.4-18.2)

" One case without data on vegetable consumption frequency.

Because subclinical HCC might be present at the time of bloodcollection and the observed elevation of serum testosterone might bedue to the presence of cancer, the association between serum testosterone level and HCC was further examined for two distinct follow-upperiods by stratifying the cases into two groups: early-onset cases who

were diagnosed within 1 year after blood collection (7 cases) andlate-onset cases who were diagnosed more than 1 year after blood

collection (28 cases). The effect of elevated testosterone on developing HCC was stronger after the exclusion of the early-onset cases

diagnosed within 1 year after blood collection. The matched RRof developing HCC for a serum testosterone level greater than5.69 ng/ml was 2.7 (95% CI = 1.04-6.80).

Table 3 shows the frequency distribution of other HCC risk factorsin liver cancer cases and controls. Liver cancer cases had a significantly higher proportion of anti-HCV-positive status (20.0%) thanmatched controls (2.9%), with a RR of 11.8 (95% CI = 2.4-57.7).

More liver cancer cases (68.6%) than matched controls (57.9%) werecigarette smokers, but the difference was not statistically significant.The percentage of habitual alcohol drinking was higher among livercancer cases (25.7%) than matched controls ( 13.6%), with a RR of 2.6(95% CI = 1.0-7.0) at the borderline significance level (0.05 < P <

0.10). There were more vegetarians who consumed mostly preservedvegetables and foodstuffs made from beans among liver cancer cases(17.1%) than among matched controls (9.3%), with a RR of 2.2 (95%CI = 0.7-6.7). Liver cancer cases had a lower percentage of fresh

vegetable consumption, defined as fewer than 6 meals/week (20.6%),than matched controls (6.4%), with a significant RR of 4.1 (95% CI= 1.3-13.4). More liver cancer cases had a history of liver diseases

diagnosed by physicians (17.1%) than matched controls (4.3%), witha RR of 5.0 (95% CI = 1.4-18.2).

Multiple conditional logistic regression analysis also showed asignificant association between elevated serum testosterone level and

liver cancer when HBsAg carrier status, anti-HCV positivity, alcohol

drinking, cigarette smoking, frequency of vegetable consumption,vegetarian habit, and past liver disease history were adjusted (Table4). The multivariate-adjusted RR for men whose serum testosteronelevels were in the upper tertile was 4.1 (95% CI = 1.3-13.2), as

compared with those having serum testosterone levels in the middle orlower tertiles (P = 0.016). The anti-HCV positivity, past liver disease

history, and low vegetable consumption frequency remained significantly associated with liver cancer. The multivariate-adjusted RR was37.0 (95% CI = 4.6-295.2) for anti-HCV positivity, 12.0 (95% CI =2.4-60.3) for past liver disease history, and 7.2 (95% CI = 1.5-33.8)

for low vegetable consumption frequency. Alcohol drinking was alsosignificantly related to HCC, with an adjusted RR of 4.3 (95% CI =1.1-16.9). Vegetarian habit was associated with liver cancer at aborderline significance level, with an adjusted RR of 3.7 (95% CI =0.9-14.6). No significant association was observed between cigarette

smoking and HCC in this study.

DISCUSSION

The striking male:female ratios in the incidence of HCC in ethnically and geographically diverse populations (1, 2) as well as thegreater susceptibility of male animals to spontaneous and chemical-induced HCC (8-14, 18) suggest that sex-related factors may be

involved in its development. The male predominance in HCC isindependent of the hereditary diversity of different races and animalspecies, as well as the varying environmental exposures in differentgeographical areas and animal experiments. It seems resonable toassume that sex hormone is primarily responsible for the genderdiscrepancy. This study shows that elevated serum levels of testosterone were associated with an increased risk of developing HCC. Theassociation remained significant after adjusting for the effect ofHBsAg carrier status, anti-HCV positivity, alcohol drinking, cigarette

smoking, past liver disease history, and dietary habits. These resultsreveal that the association between testosterone level and HCC wasindependent of other HCC risk factors. They are also compatible withthe hypothesis that testosterone has a role in the etiology of humanHCC.

The testosterone levels in this study were determined from serumsamples collected 6 months to 5 years (average, 2.4 years) before thediagnosis of HCC. To investigate whether subclinical HCC influencesthe testosterone-HCC relationship, the association between elevated

testosterone and HCC was further examined according to the intervalbetween blood collection and the diagnosis of HCC. Inasmuch as theeffect of elevated testosterone on the development of HCC was stron-

Table 4 Conditional logistic regression analysis of multiple risk factorsassociated with HCC"

VariableTestosterone

(ng/ml)Anti-HCVAlcohol drinkingCigarette smokingVegetarian habitVegetable consumption

(meals/week)Past liver disease

historyComparison>5.69

versus <5.69Positive versus negativeYes versus noYes versus noYes versusno<6

versus >6

Yes versus noMultivariate-adjusted

relative risk(95% confidenceinterval)4.1

(1.3-13.2)''37.0(4.6-295.2)'4.3(1.1-16.9)*1.2(0.4-3.1)

3.7(0.9-14.6)i/7.2(1.5-33.8)*

12.0(2.4-60.3)*'

" HBsAg carrier status was also included in the regression model to adjust for its

confounding effect.hP< 0.05.''/>< 0.001.

^ 0.05 </>< 0.1.f P<0.0\.

792

Research. on January 7, 2021. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

Page 4: Elevated Serum Testosterone Levels and Risk of Hepatocellular … · genic-anabolic steroid treatment (21) and the presence of androgen receptors in human HCC tissues (22, 23). We

SERUMTESTOSTERONI;ANDHKPATOCI-U.ULARCARCINOMA

ger after the exclusion of the early-onset cases diagnosed within 1 year

after blood collection, it is unlikely that the elevated testosteronelevels in HCC cases were a result of this cancer. However, preclinicaldisease may also have an effect on serum testosterone. Previousstudies have shown that most male patients with chronic liver diseasehad clinical and biochemical evidence of hypogonadism, and theirserum testosterone levels fell further as the disease progressed (24).One study also reported that patients of alcoholic cirrhosis combinedwith HCC had significantly lower levels of circulating testosteronethan patients affected with alcoholic cirrhosis alone (26). HCC usuallyoccurs in cirrhotic livers (27). Chronic hepatitis, liver cirrhosis, andHCC may be successive sequelae of chronic HBV infection (28). Inthis study, we cannot exclude the possibility that underlying cirrhosishas an effect on testosterone levels in the preclinical phase of someHCC cases and thus leads to an underestimation of the relative risk ofHCC associated with elevated testosterone levels. In other words, therelative risk estimate in this study was obtained under conservativecircumstances. After HB„Agcarrier status and past liver disease history were adjusted for by the conditional logistic regression analysis,an increase in RR associated with elevated testosterone levels wasobserved.

It has been well established that testosterone plays a prominent rolein mouse hepatocarcinogenesis (8, 10-12, 14, 19), but the importance

of this male hormone in the development of HCC was reported lessfrequently in rats (15, 18). In animal models, male castration decreased the incidence of chemically induced HCC (10, 12, 15, 18. 19),while chronic administration of testosterone to female or castratedmale animals increased the growth of preneoplastic hepatic foci (11)or the risk of spontaneous (8) or chemically induced HCC (18). Theseresults suggest that testosterone may promote the development ofHCC, even at physiological levels. This is consistent with our datasuggesting that men with elevated testosterone levels are at an increased risk for HCC. The mechanism of testosterone involvement inthe genesis of HCC remains to be elucidated. However, in the light ofanimal studies, this may be related to the ability of testosterone topromote the growth of preneoplastic hepatic lesions (11, 19). Themarked sex difference in susceptibility to chemically induced hepatocarcinogenesis may also be related to the sexual dimorphism in themetabolism of environmental carcinogens (12, 20, 29-31). The path

way leading to the active metabolite of aflatoxin was reported to bemore active in male animals than in females (29). Recent animalstudies also indicated that male hormone may mediate the activity ofsulfotransferase (30) and the expression of cytochrome p450 genes(31). However, the role of testosterone in carcinogen metabolism inhumans remains to be explored. A HBV transgenic mouse model alsosuggested that sex hormones may play a role in HBV gene expression(32). HBV-associated chronic hepatitis (6) and liver cirrhosis (7) are

found more frequently and may have a more aggressive behavior inmen than in women (28). It would be worthwhile to determinewhether the effect of testosterone on the development of HCC occursthrough the regulation of HBV gene expression, which causes a higherincidence of chronic hepatitis and liver cirrhosis in males than infemales.

A significantly higher proportion of serum testosterone levels above5.69 ng/ml was also observed among HBsAg-positive controls thanHBsAg-negative controls in this study. There was no established

mechanism to explain this phenomenon. The higher testosterone levels in HB„Agcarriers than in noncarriers may result from someunexplored factors which were associated with serum testosteronelevels and distributed differently between the two groups. It might alsobe possible that chronic HBV infection itself increases the circulating

testosterone level. Whether increased testosterone production is alsoone of the mechanisms for HBV-relatcd hepatocarcinogenesis remains

to be studied.Although our study suggests that the positive association between

serum testosterone level and risk of HCC observed in many animalexperiments may apply to human beings, there may be some uncontrolled factors such as genetic factors, environmental factors not studied here, or circulating hormones other than testosterone which mayalso partly result in an association between elevated testosterone andHCC risk in this study. Because testosterone is converted into estrogenand other metabolites in peripheral tissues, the circulating level oftestosterone may not completely reflect the level of testosterone functioning directly at the target site. It is worthwhile to examine theinterindividual variation in the rate of testosterone conversion intoestrogen and other metabolites and to assess the importance of thisindividual discrepancy in the induction of HCC. This is the first reporton the role of serum testosterone level in the development of HCC:further studies on other populations are needed to validate this finding.

REFERENCES

1. Editorial. Hepatocellular cancer: differences between high and low incidence regions.Lancet. 2: 1183-1184, 1987.

2. Yu. M. W., Tsai. S. F.. Hsu. K. H., You, S. L., el al. Epidemiologie characteristics ofmalignant neoplasms in Taiwan. II. Liver cancer. J. Nati. Public Health Assoc., 8:125-138. 1988.

3. Bcaslcy, R. P. Hepatitis B virus: the major etiology of hepatoccllular carcinoma.Cancer (Phila.). 61: 1942-1956. 1988.

4. Chen. C. J.. Liang. K. Y. Chang. A. S.. Chang. Y. C.. el ill. Effects of hepatitis B virus,alcohol drinking, cigarette smoking and familial tendency on hepatoccllular carcinoma. Hepatology (Ballimore). 13: 398-406. 1991.

5. Yu. M. W.. You, S. L.. Chang. A. S., Lu. S. N., Liaw. Y. F.. and Chen, C. J. Associationbetween hepatitis C virus antibodies and hepatocellular carcinoma in Taiwan. CancerRes.. 51: 5621-5625. 1991.

6. Chu. C. M.. Liaw. Y. F.. Sheen, I. S., Lin, D. Y. and Huang. M. J. Sex difference inchronic hepatitis B virus infection: an appraisal based on the status of hepatitis B eantigen and antibody. Hepatology (Baltimore), 3: 947-950. 1983.

7. Lin. T. M.. Tsu, W. T.. and Chen. C. J. Mortality of hepatoma and cirrhosis of liverin Taiwan. Br. J. Cancer. 54: 969-976. 1986.

8. Agnew. L. R. C.. and Gardner. W. L'. The incidence of spontaneous hepatomas in

C3H, C3H (low milk factor), and CBA mice and the effect of estrogen and androgenon the occurrence of these tumors in C3H mice. Cancer Res.. 12: 757-761. 1952.

9. Sawaki, M.. Enomoto. K., Takahashi, H.. Nakajima, Y, et iti. Phenotype of preneoplastic and neoplastic liver lesions during spontaneous liver carcinogcnesis of LECrats. Carcinogenesis (Lond.), //.- 1857-1861. I99().

10. Vesselinovitch, S. D.. and Mihailovich. N. The effect of gonadectomy on the development of hepatomas induced by urethan. Cancer Res., 27: 1788-1791, 1967.

11. Moore. M. R.. Drinkwater. N. R.. Miller. E. C., Miller. J. A., el al. Quantitativeanalysis of the time-dependent development of glucose-6-phosphalase-dcficient foci

in the livers of mice treated neonalally with diethylnitrosamine. Cancer Res., 41:1585-1593. 1981.

12. Toh, Y. C. Effect of neonatal castration on liver tumor induction by /V-2-tluorenylac-etamide in suckling BALB/c mice. Carcinogenesis (Lond.). 2: 1219-1221. 1981.

13. Kemp. C. J.. and Drinkwater, N. R. Genetic variation in liver tumor susceptibility,plasma testosterone levels, and androgen receptor binding in six inbred strains ofmice. Cancer Res.. 49: 5044-5047. 1989.

14. Kemp. C. J.. Leary. C. N.. and Drinkwater. N. R. Promotion of murine hepatocar-cinogencsis by testosterone is androgen receptor-dependent but not cell autonomous.Proc. Nati. Acad. Sci. USA. Hf>:7505-7509. 1989.

¡5.Firminger. H. I., and Reuber, M. D. Influence of adrenocortical. androgenic. andanabolic hormones on the development of carcinoma and cirrhosis of the liver in AXCrats fed JV-2-fluorcnyldiacetamide. J. Nati. Cancer Inst., 27: 559-595. 1961.

16. Dunsford, H. A.. Sell. S.. and Chisari, F. V. Hepatocarcinogenesis due to chronic livercell injury in hepatitis B virus transgenic mice. Cancer Res.. 50: 3400-3407, 1990.

17. Kim. C. M., Koike. K.. Saito. L.. Miyamura, T. el al. HBx gene of hepatitis B virusinduces liver cancer in transgenic mice. Nature (Lond.). 351: 317-320. 1991.

18. Reuher. M. D. Influence of hormones on /V-2-fluorenyldiacetamide-induced hyper-plastic hepatic nodules in rats. J. Nail. Cancer Inst., 43: 445-452. 1969.

19. Vesselinovitch. S. D., Ltze, L., Mihailovich. N., and Rao, K. V. N. Modifying role ofpartial hepatectomy and gonadectomy in ethylnitrosourea-induced hepatocareinogen-esis. Cancer Res.. 40: 1538-1542. 1980.

20. Goodall, C. M.. and Butler, W. H. Aflatoxin Carcinogenesis: inhibition of liver cancerinduction in hypophysectomized rats. Int. J. Cancer. 4: 422—429.1969.

21. Farrell, G. C., Joshua, D. E., Ui-en, R. F.. Baird. P. J.. el al. Androgen-induccdhepatoma. Lancet. /: 430-432. 1975.

22. Ohnishi. S.. Murakami. T.. Moriyama, T. Mitamura, K.. and Imawari, M. Androgenand estrogen receptors in hepatocellular carcinoma and in the surrounding noncan-cerous liver tissue. Hepatology (Baltimore), fi: 440-443. 1986.

793

Research. on January 7, 2021. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

Page 5: Elevated Serum Testosterone Levels and Risk of Hepatocellular … · genic-anabolic steroid treatment (21) and the presence of androgen receptors in human HCC tissues (22, 23). We

SERUM TESTOSTERONE AND HEPATOCELLULAR CARCINOMA

23. Nagasue, N., Ito, A., Yukaya. H., and Ogawa, Y. Androgen receptors in hepatocellular 29. O'Brien, K., Mose, E., Judah, D., and Neal, G. Metabolic basis of the species

carcinoma and surrounding parenchyma. Gastroenterology, 89: 643-647, 1985. difference to aflatoxin Bpinduced hepatotoxicity. Biochem. Biophys. Res. Commun.,24. Johnson, P. J. Sex hormones and the liver. Clin. Sci., 66: 369-376, 1984. 114: 813-827, 1983.25. Dai, W. S., Kuller, L. H., La Porte, R. E., Gutai, J. P., el al. The epidemiology of 30. Surh, Y. J., Liem, A., Miller, E. C., and Miller, J. A. Age- and sex-related difference

plasma testosterone levels in middle-aged men. Am. J. Epidemici., 114: 804-816, in activation of the carcinogen 7-hydroxymethyl-12-methylbenz[a]anthracene to an1981. electrophilic sulfuric acid ester metabolite in rats. Biochem. Pharmacol.. 41: 213-221,

26. Guechot, J., Peigney, N., Ballet, F., Vaubourdolle, M., et al. Sex hormone imbalance 1991.in male alcoholic cirrhotic patients with and without hepatoceilular carcinoma. Cancer 31. Henderson, C. J., and Wolf, C. R. Evidence that the androgen receptor mediates sexual(Phila.), 62: 760-762, 1988. differentiation of mouse renal cytochrome p450 expression. Biochem. J., 278: 499-

27. Johnson, P. J., and Williams, R. Cirrhosis and aetiology of hepatocellular carcinoma. 503, 1991.J. Hepatol., 4: 140-147, 1987. 32. Farza, H., Salmon, A. M., Hadchouel, M., Moreau, J. L.. el al. Hepatitis B surface

28. Beasley, R. P. Hepatitis B virus as the etiologic agent in hepatocellular carcinoma— antigen gene expression is regulated by sex steroid and glucocorticoids in transgenicepidemiologie considerations. Hepatology (Baltimore), 2: 21s-26s, 1982. mice. Proc. Nati. Acad. Sci. USA. 84: 1187-1191, 1987.

794

Research. on January 7, 2021. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from

Page 6: Elevated Serum Testosterone Levels and Risk of Hepatocellular … · genic-anabolic steroid treatment (21) and the presence of androgen receptors in human HCC tissues (22, 23). We

1993;53:790-794. Cancer Res   Ming-Whei Yu and Chien-Jen Chen  CarcinomaElevated Serum Testosterone Levels and Risk of Hepatocellular

  Updated version

  http://cancerres.aacrjournals.org/content/53/4/790

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  [email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/53/4/790To request permission to re-use all or part of this article, use this link

Research. on January 7, 2021. © 1993 American Association for Cancercancerres.aacrjournals.org Downloaded from