REPORT Nonviral Risk Factors for Hepatocellular Carcinoma in a Low-Risk Population, the Non-Asians of Los Angeles County, California

We conducted interviews on 74 patients with histologically confirmed hepatocellular carcinoma. These patients, aged 18-74 years, were black or white residents of Los Angeles County. We also interviewed 162 population control subjects who were comparable to the case patients by age, sex, and race. Cigarette smoking was a significant risk factor for hepatocellular carcinoma [relative risk (RR) = 2.1; 95% confidence limits (CL) = 1.1, 4.0]; the effects were similar in men and in women. Heavy alcohol consumption was another risk factor for hepatocellular carcinoma in men; men who consumed 80 g or more of ethanol per day had an RR of 4.7 (95% CL = 1.4, 15.4) relative to those who had never drunk alcohol on a weekly basis. The level of alcohol intake was relatively low in women, and no significant effect on risk of hepatocellular carcinoma was observed. Use of oral contraceptives was significantly related to risk of hepatocellular carcinoma in women (RR = 3.0; 95% CL = 1.0, 8.8); those who were exposed for more than 5 years exhibited a 5.5-fold increased risk (95% CL = 1.2, 24.8). The effects of these three risk factors on hepatocellular carcinoma development were independent of each other and independent of serologically determined viral hepa1820

Hepatocellular carcinoma is a rare malignancy among non-Asians in the United States. In Los Angeles County, the average annual age-standardized (world) incidence rates for primary liver cancer during 1978 through 1982 in Hispanic white, black, and non-Hispanic white men were 4.7, 4.5, and 2.1 per 100 000, respectively. The corresponding rates in women were 1.6, 1.1, and 0.9 per 100 000 (/). Chronic infection with the hepatitis B virus (HBV) is a well-established causal factor in the development of hepatocellular carcinoma (2). Among populations at high risk for hepatocellular carcinoma, such as the southern Chinese, virtually all case patients with hepatocellular carcinoma were HBV carriers (3,4). Among U.S. blacks and whites, however, only half of all hepatocellular carcinoma cases in men and fewer than one third of cases in women could be attributable to a viral etiology (5,6). Cigarette smoking and alcohol drinking have been suggested as major nonviral risk factors for hepatocellular carcinoma (7-10). In 1984, we initiated a case-control study among blacks and whites of Los Angeles County to investigate the possible association between tobacco and alcohol use and the development of hepatocellular carcinoma. Other factors of interest were hormone use, medical history,

history of blood transfusion and other potential exposures to contaminated blood, occupational exposure to chemicals, and exposure to nitrosamine-rich foods. The results of this study to date are described in this report.

Subjects and Methods We studied histologically confirmed cases of newly diagnosed hepatocellular carcinoma in black and white residents of Los Angeles County who were between the ages of 18 and 74 years. Cases were identified through the Los Angeles County Cancer Surveillance Program (77), a population-based cancer registry that records all cases of cancer microscopically verified or recorded on a death certificate. Cases identified between January 1984 and February 1990 were eligible for inclusion in the study, which consisted of a personal interview and, on consent, collection of a venous blood specimen. There were 412 eligible case patients; 291 died prior to our attempted contact, 11 could not be located, and in 26 instances, the physician or patient refused to cooperate. The remaining 84 case patients

Received April 2, 1991; revised September 11, 1991; accepted September 24, 1991. Supported by grant SIG-2 from the American Cancer Society and Public Health Service grant CA17054 from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services. M. C. Yu, B. E. Henderson, Department of Preventive Medicine, University of Southern California School of Medicine, Los Angeles, Calif. M. J. Tong, Liver Center, Huntington Memorial Hospital, Pasadena, Calif. S. Govindarajan, Liver Unit, University of Southern California School of Medicine, Los Angeles. We thank Mrs. Susan Roberts for conducting interviews on and collecting blood specimens from study subjects and Mrs. Kazuko Arakawa for computing assistance. 'Correspondence to: Mimi C. Yu, Ph.D., Kenneth Norris Jr. Comprehensive Cancer Center, University of Southern California, 1441 Eastlake Ave., Los Angeles, CA 90033-0800.

Journal of the National Cancer Institute

Downloaded from http://jnci.oxfordjournals.org/ at Freie Universitaet Berlin on May 7, 2015

Mimi C. Yu* Myron J. Tong, Sugantha Govindarajan, Brian E. Henderson

titis. Our data suggest that cigarette smoking, alcohol consumption, and use of oral contraceptives are major risk factors for hepatocellular carcinoma among non-Asian residents of Los Angeles County. We also observed a significant association between a history of diabetes and hepatocellular carcinoma (RR = 33; 95% CL = 1.5, 7.2), especially among those who had received insulin treatment (RR = 18.5; 95% CL = 2.2, 156.0). This association may have etiological significance. [J Natl Cancer Inst 83:1820-1826,1991]

Vol. 83, No. 24, December 18, 1991

(87%) were bom in the United States. Level of education was comparable between cases and controls (P - .20). Relative to non-tobacco users, there was no increase in risk for hepatocellular carcinoma among users of cigar, pipe, or chewing tobacco only [relative risk (RR) = 0.8; 95% confidence limits (CL) = 0.2, 3.5]. On the other hand, cigarette smokers were twice as likely to develop hepatocellular carcinoma relative to non-tobacco users (RR = 2.0; 95% CL = 1.0, 4.1); this positive association was present in both men (RR = 1.9) and women (RR = 2.1). Table 1 presents the pattern of cigarette consumption among hepatocellular carcinoma case patients and control subjects. The RR for hepatocellular carcinoma was lower in former cigarette smokers (RR = 1.6) than in current smokers (RR = 2.5). Among ex-smokers, risk declined with increasing number of years since quitting. Among current smokers, there was no indication that risk increased with increasing number of cigarettes smoked each day. Cigarette smoking remained significantly associated with risk of hepatocellular carcinoma after adjustment for total drink-years (defined below) of alcohol consumption (Table 1). Table 1 also lists the pattern of alcohol consumption among hepatocellular carcinoma case patients and control subjects. A drinker was defined as someone who consumed any alcoholic beverage at least once a week. We further classified drinkers by the usual amount of total ethanol he/she consumed each day and by a measure of cumulative alcohol intake defined as the cross-product of the number of drinks the subject usually consumed each day and the number of years he/she had been maintaining that habit of drinking. (A drink is equivalent to one can of beer, or one glass of wine, or one shot of hard liquor.) Among men, there was a 1.9-fold increase in risk for hepatocellular carcinoma among drinkers relative to non-drinkers (95% CL = 0.8, 4.5). There was a significant dose-response relationship between risk and exposure, regardless of whether the exposure index was daily ethanol intake (linear trend P = .003) or cumulative drink-years (linear trend P = .003). Alcohol intake remained significantly associated with hepatocellular carcinoma in men after adjustment for cigarette use (Table 1). Very few

REPORT 1821

Downloaded from http://jnci.oxfordjournals.org/ at Freie Universitaet Berlin on May 7, 2015

(20% of all eligible patients) were inter- formed by hospitals/physicians on anviewed, and pathology slides from each other nine case patients, and their results case patient were subsequently reviewed had been made available to us. Therefore, by a hepatopathologist (most were re- HBV serology was known for 71 (96%) viewed by S. Govindarajan). Eight case of the 74 cases. Serum samples from 51 patients with metastatic cancer and two case patients were also tested for the prescase patients with liver adenoma were ex- ence of antibodies to hepatitis C virus (anticluded. Thus, 74 hepatocellular carcinoma HCV), using standardized kits from Ortho Diagnostic Systems (Raritan, N. J.) (5). patients were included in the study. For each of the 84 interviewed case Three parallel analyses were performed patients, we sought to recruit two control on the study data. We used the cross-classubjects from the neighborhood where the sification and unconditional logistic recase patient resided at the time of diag- gression methods (72) to analyze the data nosis. Using the house of each case pa- on the 74 case patients and 162 control tient as a reference point and proceeding subjects, stratifying by age, sex, and race. in a systematic and invariable sequence, We also used the matched pairs and conwe canvassed up to 150 residential units ditional logistic regression methods (72) to identify two control subjects who could to examine the 74 matched sets (with 144 be matched to the case patients by sex, control subjects, of which 18 were not year of birth (within 5 years), and race matched to the index case patient by race) (Hispanic white, non-Hispanic white, black). and their subset of 68 matched sets, We attempted to identify the sex, age, and whose 126 control subjects were all race of the residents of each housing unit. matched to the index case by race. Our goal was to interview the first two Results from the three parallel analyses residents (in separate households) in each were similar to each other. All statistics sequence who met the matching criteria. presented in this report were based on the If a potential control subject refused, the stratified analysis of the 74 case patients second eligible control subject in the se- and 162 control subjects. Patients or subquence would be asked to participate. We jects who had a missing value were relaxed the race-matching criterion if we eliminated from the analysis of the corfailed to recruit two control subjects responding variable. The unconditional within 150 housing units. logistic regression method (72) was used A total of 162 control subjects were to examine exposures within viral hepatirecruited into the study; most (91%) were tis-positive and -negative case patients the first (71%) or the second (20%) separately, with all of the control subjects eligible neighbors. One hundred forty- as the comparison group. All P values four control subjects were matched to the quoted were two-sided. Total ethanol in74 hepatocellular carcinoma patients. Eigh- take was calculated using conversion facteen control subjects were not matched to tors of 12.96, 12.93, and 14.03 g of ethanol for each serving of beer (one can the index case by race. Except for two control subjects who or 12 fl. oz.), wine (one glass or 3.5 fl. preferred to be interviewed on the tele- oz.), and hard liquor (one shot or 1.5 fl. phone, all study subjects were inter- oz.), respectively (13). viewed in person in their homes by an experienced interviewer using a structured Results questionnaire that asked about past events occurring 2 or more years prior to the Forty-nine (66%) case patients and 104 case patient's diagnosis of hepatocellular (64%) control subjects were men. Eighcarcinoma. A blood specimen was ob- teen (24%) case patients were Hispanic tained from 62 (84%) of the 74 case white, 39 (53%) were non-Hispanic white, patients in the study. All serum samples and the remaining 17 (23%) were black. were tested for hepatitis B surface antigen Among control subjects, 36 (22%) were (HBsAg), antibodies to the hepatitis B Hispanic white, 96 (59%) were noncore antigen (anti-HBc), and antibodies to Hispanic white, and the remaining 30 HBsAg (anti-HBs) by radioimmunoassay, (19%) were black. The mean ages for using standardized kits from Abbot case patients and control subjects were Laboratories (North Chicago, 111.). These 59.6 and 58.8 years, respectively. Most serological determinations had been per- case patients (85%) and control subjects

Table 1. Use of tobacco and alcohol among hepatocellular carcinoma case patients and control subjects

Ca/Co* Cigarettes Never smoker 19/66 22/48 Ex-smoker 13/36 Quit 10+y ago Quit 1-9 y ago 9/12 Current smoker 33/48 Up to 19 cigarettes • 15/18 per day 20+ cigarettes 18/30 per day Adjusted for alcohol drinking! Ex-smoker who quit 13/36 10+ years ago Recent/current smoker 42/60

Adjusted for cigarette uset Nondrinker/29 drink-y or under 30-59 drink-y 60+ drink-y

Ca/Co*

Males RR

95% CL

Ca/Co*

Females RR

95% CL

0.3, 6.5 0.1,8.0 0.3, 15.0 0.8, 6.9 0.6, 7.7

1.0 1.6 1.2 2.6 2.5 2.8

0.7, 3.5 0.5, 3.0 0.9,7.4 1.2,5.0 1.2,6.9

8/30 19/41 12/32 7/9 22/33 9/9

1.0 1.8 1.4 3.0 2.8 3.7

0.7, 4.8 0.5,4.1 0.8, 10.8 1.0,7.9 1.0, 13.3

11/15 6/9

1.0 1.4 0.8 2.1 2.4 2.1

2.2

1.0,5.0

13/24

2.4

0.8,7.4

5/6

2.9

0.7, 11.7

1.1

0.4, 2.6

12/32

1.1

0.4, 3.3

1/4

0.8

0.1,8.9

2.1

1.1,4.3

29/42

2.2

0.8,6.0

13/18

2.4

0.9,6.7

25/68 13/45

1.0 0.8

0.4, 1.8

9/32 9/32

1.0 1.0

0.3,2.8

16/36 4/13

1.0 0.7

0.2, 2.6

10/20 26/29

1.5 2.7

0.6, 3.6 1.2,5.9

7/13 24/27

1.9 3.1

0.6,6.2 1.2,7.8

3/7 2/2

1.0 2.1

0.2, 4.8 0.2, 17.5

38/113

1.0

18/64

1.0

20/49

1.0

10/20 26/29

1.4 2.4

7/13 24/27

1.9 2.7

3/7 2/2

0.9 1.8

0.6, 3.4 1.2,5.0

0.7, 5.7 1.2,5.9

11/36 3/7 1/4

2/3

0.2, 4.3 0.2, 15.2

*No. of case patients/No, of control subjects. t Adjusted for total drink-y of alcohol consumption. ^Adjusted for three smoking categories as follows: non-user, quit 10+ y ago, recent or current user.

women were heavy drinkers, and there was no significant association between risk and either daily ethanol intake (linear trend P = .34) or cumulative drink-years (linear trend P = .30). The relationship between risk of hepatocellular carcinoma and use of cigarettes and alcohol in men is further analyzed in Table 2. Increased risks were observed among men who consumed cigarettes but not alcohol, or vice versa, and risk was higher when both factors were present Table 3 presents the adjusted RRs for alcohol drinking and cigarette smoking within HBV/HCV-positive and -negative cases of hepatocellular carcinoma separately. The observed RRs in the two groups of patients suggested an additive interaction effect between cigarette smoking (or alcohol drinking) and viral hepatitis on risk of hepatocellular carcinoma. Use of oral contraceptives was significantly associated with risk of hepatocellular carcinoma in women (RR = 3.0; 95% CL = 1.0, 8.8) (Table 4). There was increasing risk with increasing duration of use (linear trend P = .03), and the RR for 5 or more years of use was 5.5 (95% CL = 1.2, 24.8). When we restricted

1822

the analysis to women aged 64 years and younger, i.e., those who were in their prime reproductive years when oral contraceptives were first available in the early 1960s, the association became stronger. Among these younger women, there were

13 case patients and 16 control subjects who had used oral contraceptives (RR = 4.9; 95% CL = 1.4, 17.8), and the RR for 5 or more years of use was 33.2 (95% CL = 2.8, 390.7). Most of the users were past users (11 case patients and 16 control

Table 2. Relative risks for hepatocellular carcinoma in men according to their levels of cigarette smoking and alcohol drinking

Cigarette—never Alcohol Never Ever Cigarette—ever Alcohol Never Ever Cigarette—nonsmoker Alcohol 29 drink-y or under 30+ drink-y Cigarette—ex-smoker Alcohol 29 drink-y or under 30+drink-y Cigarette—current smoker Alcohol 29 drink-y or under 30+drink-y

Ca/Co*

RR

95% CL

3/14 5/16

1.0 1.4

0.3, 7.0

6/18 35/56

1.6 3.0

0.3, 7.7 0.8, 11.8

4/24 4/6

1.0 4.2

0.8, 22,2

5/23 14/18

1.4 4.8

0.3,6.0 1.3, 17.4

9/17 13/16

3.7 5.4

0.9, 15.5 1.4,21.0

*No. of case patients/No, of control subjects.

Journal of the National Cancer Institute

Downloaded from http://jnci.oxfordjournals.org/ at Freie Universitaet Berlin on May 7, 2015

Alcohol Non drinker 29 drink-y or under 30-59 drink-y 60+drink-y

Total No. of subjects RR 95% CL

Table 3. Interaction of cigarette and alcohol use and HBV/HCV infection on risk of hepatocellular carcinoma in men HBV/HCV+''cases 95% CL Adjusted! RR

Alcohol Nondrinker/29 drink-y or under 30-59 drink-y 60+drink-y Cigarette use Nonuser/long-term quitter Recent/current smoker

HBV/HCV-cases 95% CL Adjusted! RR

1.0 0.9 1.3*

0.1,6.0 0.3,4.6

1.0 1.8 6.1

0.2, 15.2 1.6.23.2

1.0 1.8§

0-5,6.2

1.0 3.4

0.9, 12.5

*Presence of one or more of the following serological markers: HBsAg, anti-HBs, anti-HBc, anti-HCV. RR (HBV/HCV+) = 7.2. fEstimated from unconditional logistic regression model, which included age, race, history of hepatitis/cirrhosis, history of diabetes, alcohol drinking, and cigarette smoking. •Under an additive model, RR (HBV/HCV+ and 60+ drink-y) = 7.2 + 6.1 - 1 = 12.3. Therefore, predicted RR (60+ drink-y given HBV/HCV+) = 12.3/7.2 = 1.7. §Under an additive model, RR (HBV/HCV+ and recent/current smoker) = 7.2 + 3.4 - 1 = 9.6. Therefore, predicted RR (recent/current smoker given HBV/HCV+) = 9.6/7.2 = 1.3.

Ca/Co* Oral contraceptives No Yes 12 moor under 13-60 mo 61+mo Premarin No Yes Up to 12 mo 13-60 mo 61+mo

RR (95% CL)

Adjusted RR 95% CL)

12/40 13/18 4/7 3/7 6/4

1.0 3.0 2.2 1.8 5.5

(1.0,8.8) (0.5, 10.0) (0.3,9.3) (1.2,24.8)

l.Ot 3.0 2.3 1.7 5.5

(1.0,9.0) (0.5, 10.9) (0.3,9.1) (1.2,24.8)

15/39 10/19 3/7 3/6 3/6

1.0 1.5 1.1 1.3 1.4

(0.5,4.4) (0.2,5.4) (0.3,6.3) (0.3,7.6)

1.0* 1.1 0.8 1.0 1.0

(0.3,3.6) (0.2,4.5) (0.2,5.1) (0.2,6.0)

•No. of case patients/No, of control subjects. t Adjusted for duration of use of Premarin/other estrogens. t Adjusted for duration of use of oral contraceptives.

subjects), and there were no detectable differences in risk by the number of years since first or last use, respectively. (A past user was defined as someone who had stopped using oral contraceptives at the time of cancer diagnosis of the index case.) Among the 13 case patients who had used oral contraceptives, the mean duration since first use was 17.7 years; for the 11 former users, the mean duration since last use was 14.5 years. Seven case patients and six control subjects provided information on specific oral contraceptives used. All were combination-type oral contraceptives. Six case patients and five control subjects were exposed to mestranol-containing formulations, while one case patient and one control subject were exposed to ethinyl estradiol-containing formulations.

Vol. 83, No. 24, December 18, 1991

Use of Premarin or other estrogens was positively, though not significantly, related to hepatocellular carcinoma development (RR = 1.5; 95% CL = 0.5, 4.4) (Table 4). There was a nonsignificant increase in risk by duration of use (linear trend P = .62). The association between hepatocellular carcinoma and Premarin/ other estrogen use was reduced after adjustment for duration of use of oral contraceptives. In contrast, the elevated risks associated with use of oral contraceptives remained practically unchanged after adjustment for duration of use of Premarin/other estrogens (Table 4). Seven female patients with hepatocellular carcinoma were positive for one or more serological markers of HBV and HCV infections; three case patients were users of oral contraceptives. When we ex-

REPORT 1823

Downloaded from http://jnci.oxfordjournals.org/ at Freie Universitaet Berlin on May 7, 2015

Table 4. Use of oral contraceptives and Premarin/other estrogens among female hepatocellular carcinoma case patients and control subjects

cluded these seven HBV/HCV-positive cases from the analysis, the association between oral contraceptive use and hepatocellular carcinoma risk became stronger, while Premarin use remained unrelated to hepatocelluJar carcinoma after adjustment for duration of use of oral contraceptives. We also examined the relationship between use of exogenous hormones and hepatocellular carcinoma risk in women while adjusting for consumption of cigarettes and alcohol. Results were similar to those without such adjustments. Subjects were asked if they had ever been told by a doctor that they had any of a list of medical conditions; if so, the year in which the condition was first diagnosed and any treatments for it were noted. A history of medically diagnosed hepatitis (18 case patients and four control subjects; RR = 13.9; 95% CL = 4.4, 43.4) and cirrhosis of the liver (12 case patients and no control subjects; lower 95% CL = 7.7) were both significantly associated with the development of hepatocellular carcinoma. A history of blood transfusion was also positively related to hepatocellular carcinoma (RR = 2.1; 95% CL = 1.0, 4.2), as were other potential exposures to contaminated blood: male homosexuality (two case patients and three control subjects), and use of illicit drugs (three case patients and no control subjects). Diabetics were at a significantly increased risk for hepatocellular carcinoma (18 case patients and 14 control subjects; RR = 3.3; 95% CL = 1.5, 7.2), and risk was higher for those diagnosed 10 or more years ago (10 case patients and six control subjects; RR = 4.3; 95% CL = 1.5, 12.5) and those who had received insulin treatment (seven case patients and one control subject; RR - 18.5; 95% CL = 2.2, 156.0). The mean age at onset among diabetic case patients was comparable to that in diabetic control subjects (50.6 years in case patients versus 53.4 years in control subjects). Diabetes is a possible complication of cirrhosis and chronic hepatitis (14,15). Indeed, of the seven case patients and one control subject who reported both a history of hepatitis/cirrhosis and diabetes, in all instances the diagnosis of diabetes occurred either considerably later (three case patients and one control subject) or around the time the diagnosis of hepatitis or cir-

Discussion A number of case-control studies have investigated the possible association between alcohol and cigarette use and the development of hepatocellular carcinoma, and the results have been mixed. Some studies have found smoking and drinking to be independent risk factors for hepatocellular carcinoma {8-10,16-18), others observed an independent effect for alcohol intake (19-24) or cigarette smoking (725) only, while a few have failed to detect a significant association for either exposures (2627). Kew et al. (25) examined cigarette smoking among black hepatocellular carcinoma patients in South Africa and did not notice a dose-response relationship, although nonsignificant increased risks were observed among patients without any serum markers of HBV infection. Recently, Hsing et al. (29) observed a statistically significant and dose-dependent association between cigarette smoking and liver cancer mor-

1824

tality among the nearly 250 000 U.S. veterans who had been followed prospectively for 26 years. Information on alcohol use was not collected in that cohort study. Because of the rapidly fatal nature of hepatocellular carcinoma, most studies mentioned above (7,8,16,19-2123-28) were hospital based and employed other hospitalized patients as control subjects. It is recognized that cigarette smoking and, to a lesser extent, alcohol consumption are associated with illness (and, therefore, hospitalization) in general (3032), thus raising the possibility of biased comparison groups in these studies. The liver is a frequent site of metastasis. Many of the previous studies included a high proportion (greater than 40%) of case patients without histological confirmation (7,17,18212325), and none conducted an independent pathology review by an experienced hepatopathologisL Proxy interviews with family members were permissible in some studies (10,1922), while one other obtained information via mailed questionnaires (17); the quality of information obtained by these means is likely to be compromised. Any one or a combination of these design considerations could contribute to the current confusing state of knowledge regarding use of cigarette and alcohol and the development of hepatocellular carcinoma. In the present study, we attempted to avoid the design flaws described above. The control subjects were healthy individuals chosen from the neighborhood where hepatocellular carcinoma case patients resided at the time of their cancer diagnosis. The pathology slides from all case patients were reviewed by an experienced hepatopathologist to rule out metastatic cancers. All except two subjects were interviewed in person, and questions relating to alcohol consumption were designed specifically to minimize possible underreporting by heavy drinkers (the interviewer used a chart to facilitate detailed recording of intake of beer, wine, and cocktail before, during, and after each meal on each day of a typical week). Serological markers of HBV infection were known in 96% of the case patients, thus allowing for the control of this strong confounder. Our study is not without its methodological shortcomings. Because of the

rapidly fatal nature of the disease, a large proportion of hepatocellular carcinoma patients eligible for entry in our study had died by the time we identified them. Consequently, our findings may have reflected the survival advantage of subgroups of patients instead of causality. We have compared the distributions of all demographic variables routinely collected by the Los Angeles County Cancer Surveillance Program between studied and eligible cases of hepatocellular carcinoma in men and women separately. Rates of participation were not uniform across groups defined by age or marital status at the time of cancer diagnosis. We therefore examined the effects of the various risk factors in the different subgroups of case patients with dissimilar rates of participation. Results of all of these subgroup analyses were remarkably similar to the results based on all studied case patients and control subjects. Our data were supportive of use of alcohol and cigarettes as independent risk factors for hepatocellular carcinoma. They further confirmed our earlier observations (8-10) that cigarette use and HBV interact in an additive manner in asserting their simultaneous effect on risk of hepatocellular carcinoma. The observed simultaneous effect of alcohol consumption and viral hepatitis on risk of hepatocellular carcinoma also seems to be compatible with the expected effect based on an additive model (Table 3). Heavy alcohol consumption has long been postulated as a risk factor for hepatocellular carcinoma because of its hepatotoxic effects and its relationship to cirrhosis (33). Less certain has been the possible independent role of cigarette smoking in hepatocarcinogenesis. Because of the close correlation between alcohol drinking and cigarette smoking, there has been speculation that the apparent independent effect of cigarette smoking observed in some case-control studies is merely the residual effect of alcohol consumption as a result of underreporting of exposure by heavy drinkers. Our study has considerably strengthened the argument for cigarette smoking as an independent risk factor for hepatocellular carcinoma. Among men, we noted an increased risk for cigarette smoking in nonconsumers of alcohol. Too few subjects belonged in this category to allow for a

Journal of the National Cancer Institute

Downloaded from http://jnci.oxfordjournals.org/ at Freie Universitaet Berlin on May 7, 2015

rhosis was made. When we restricted our study subjects to either those who did not have a history of chronic liver disease or those who did not have any serological markers of HBV or HCV infection, the positive association between diabetes and hepatocellular carcinoma persisted. Again, risk was higher among those diagnosed in the more distant past and among those who had received insulin treatment. The multivariate logistic regression method was used to examine multiple risk factors for hepatocellular carcinoma simultaneously in men and in women separately. For men, the risk factors were use of cigarettes and alcohol and histories of hepatitis/cirrhosis and diabetes; for women, they were use of cigarettes and oral contraceptives and histories of hepatitis/cirrhosis and diabetes. All risk factors remained positively related to risk of hepatocellular carcinoma after adjustments for each other. We also examined the simultaneous effects of these risk factors among case patients with no serological markers of HBV or HCV infection (23 male and 18 female case patients). For the latter analysis, the magnitude of all adjusted RRs except those associated with a history of cirrhosis/hepatitis exceeded the corresponding RRs based on all cases.

Vol. 83, No. 24, December 18, 1991

Three previous studies had reported an excess of diabetics among hepatocellular carcinoma patients (14,46,47). La Vecchia et al. (46) observed that the excess of diabetics was especially pronounced among younger patients with hepatocellular carcinoma (under 60 years of age versus 60 or older). Our data support their findings. The RR for a history of diabetes was 7.0 (95% CL = 1.2, 41.3) in cases under 60 years of age at diagnosis; the comparable RR in those diagnosed at an older age was 2.7 (95% CL = 1.1, 6.5). Lawson et al. (47) pointed out that risk was higher among diabetics who had received drug treatment (either insulin or oral hypoglycemic agents). We did not observe any difference in risk between diabetics on diet control alone and those on hypoglycemic agents alone; both risks were about twofold to threefold. We did, however, note a huge excess of insulindependent diabetics among our case patients. Diabetes is a possible complication of chronic liver disease (14,15), and our data indicated that diabetes was secondary to hepatitis/cirrhosis among those subjects who reported a history of both conditions. However, we also noted a very strong association between diabetes and hepatocellular carcinoma among case patients who were serologically negative for HBV/HCV and did not report a history of hepatitis and/or cirrhosis (RR = 7.6; 95% CL = 2.7, 21.3). It is possible that the association in the latter group reflects the presence of subclinical cases of hepatitis/cirrhosis. An alternative explanation is that the diabetics represent a distinct etiologic subgroup of hepatocellular carcinoma patients. An as yet unidentified agent could be the cause of both the diabetes, which was likely to be insulin dependent, and hepatocellular carcinoma among these case patients. Despite the relatively small sample size and low participation rate of our study, computation of population attributable risks based on the current data set can still be useful in providing a crude estimate of the extent to which nonviral environmental exposures influence the occurrence of hepatocellular carcinoma in black and white residents of Los Angeles County. According to such computations, 56% of hepatocellular carcinoma cases in nonAsian men in Los Angeles could be attributed to cigarette smoking and/or

alcohol drinking (17% to smoking alone, 11% to drinking alone, and 28% to both smoking and drinking). In women, 54% of our cases could be attributed to cigarette smoking and/or use of oral contraceptives (16% to smoking alone, 25% to use of oral contraceptives alone, and 13% to both smoking and oral contraceptive use).

References (/) MUIR C, WATERHOUSE J, MACK TM, ET AL:

Cancer Incidence in Five Continents. V. IARC SciPubl88, 1987 (2) BLUMBERG BS, LONDON WT: Hepatitis B virus

and the prevention of primary cancer of the liver. JNCI 74:267-273, 1985 (3) BEASLEY RP, HWANG LY: Epidemiology of

hepatocellular carcinoma. In Viral Hepatitis and Liver Disease (Vyas GN, Dienstag JL, Hoofhagle JH, eds). New York: Grune & Stratton, 1984, pp 209-224 (4) YEH FS, YU MC, MO CC, ET AL: Hepatitis B virus, aflatoxins, and hepatocellular carcinoma in southern Guangxi, China. Cancer Res 49:2506-2509, 1989 (5) YU MC, TONG MJ, COURSAGET P, ET AL: Prevalence of hepatitis B and C viral markers in black and white patients with hepatocellular carcinoma in the United States. J Natl Cancer Inst 82:1038-1041, 1990 (6) YARRISH RL, WERNER BG, BLUMBERG BS: As-

sociation of hepatitis B virus infection with hepatocellular carcinoma in American patients. Int J Cancer 26:711 -715, 1980 (7) TRICHOPOULOS D, MACMAHON B, SPARROS L,

ET AL: Smoking and hepatitis B-negative primary hepatocellular carcinoma. JNCI 65:111-114,1980 (8) LAM KC, YU MC, LEUNG JWC, ET AL:

Hepatitis B virus and cigarette smoking: Risk factors for hepatocellular carcinoma in Hong Kong. Cancer Res 42:5246-5248, 1982 (9) Yu MC, HENDERSON BE: A case-control study of hepatocellular carcinoma and the hepatitis B virus, cigarette smoking, and alcohol consumption. Cancer Res 47:654-655, 1987 (10) Yu MC, MACK

TM, HANISCH R, ET AL:

Hepatitis, alcohol consumption, cigarette smoking, and hepatocellular carcinoma in Los Angeles. Cancer Res 43:6077-6079, 1983 (//) MACK MT: Cancer surveillance program in Los Angeles County. Natl Cancer Inst Monogr 47:99-101,1977 (12) BRESLOW NE, DAY NE: Statistical Methods in Cancer Research. The Analysis of Case-Control Studies. I. IARC Sci Publ 32, 1980 (13) ADAMS CF: Nutritive value of American foods in common units. Agriculture Handbook No. 456, United States Department of Agriculture. Washington, DC: US Govt Print Off, 1975, p 31 (14) KINGSTON ME, Au MA, ATIYEH M, ET AL:

Diabetes mellitus in chronic active hepatitis and cirrhosis. Gastroenterology 87:688-694, 1984 (15) DEL VECCHIO BLANCO C, GENTILE S, MARMO

R, ET AU Alterations of glucose metabolism in chronic liver disease. Diabetes Res Clin Pract 8:29-36,1990 (16) Yu

H,

HARRIS

RE,

KABAT GC,

ET AL:

Cigarette smoking, alcohol consumption, and

REPORT 1825

Downloaded from http://jnci.oxfordjournals.org/ at Freie Universitaet Berlin on May 7, 2015

smoking status-risk assessment. However, among male non-drinkers and light drinkers (in whom no elevation in risk was observed), there was a clear risk: gradient among ex-smokers and current smokers, respectively (Table 2). Furthermore, cigarette smoking was a risk factor for hepatocellular carcinoma among women in our study population in whom alcohol consumption was relatively low and no significant association with hepatocellular carcinoma was observed. Finally, experimental data have lent support to the notion that hepatocytes are susceptible to the carcinogenic effects of tobacco. Rodents exposed to tobacco or its constituents showed a significantly higher occurrence of hepatocellular carcinoma than control animals (34-36). Estrogens, including those used in oral contraceptives, are powerful promoters of hepatocarcinogenesis in animals (37,38), and seven case-control studies have investigated the possible association between oral contraceptive use and hepatocellular carcinoma development (39-45). Five of these studies were conducted among white women with a low incidence of hepatocellular carcinoma (39-43). Despite the small sample sizes and various other methodological limitations, these studies consistently reported a strong association between use of oral contraceptives, especially long-term use, and risk of hepatocellular carcinoma. The two studies conducted among women in regions endemic for HBV infection and hepatocellular carcinoma, on the other hand, did not demonstrate any increase in risk for hepatocellular carcinoma among users of oral contraceptives relative to non-users (44,45). The latter observation suggests that the interactive effect of oral contraceptive use and HBV infection is additive rather than multiplicative. Under the assumption of additive risks, a sample size many times larger than sample sizes in the World Health Organization Collaborative Study (44) and the South African Study (45) would be required to detect the additional risk in oral contraceptive users against the very high background risk in HBV carriers. Our study is not large enough to shed light on whether the interactive risk of oral contraceptive use and HBV infection on hepatocellular carcinoma is additive or multiplicative. Our data are compatible with both models.

primary liver cancer: A case-control study in the USA. Int J Cancer 42:325-328, 1988 (17) OSHIMA A, TSUKUMA H, HjYAMA T, ET AL: Follow-up study of HBsAg-positive blood donors with special reference to effect of drinking and smoking on development of liver cancer. Int J Cancer 34:775-779, 1984 (18) TANAKA K, HIROHATA T, TAKESHITA S, ET AL:

Blood transfusion, alcohol consumption, and cigarette smoking in causation of hepatocellular carcinoma: A case-control study in Fukuoka, Japan. Jpn J Cancer Res 79:10751082,1988 (19) STEMHAGEN A, SLADE J, ALTMAN R, ET AL:

Occupational risk factors and liver cancer. A retrospective case-control study of primary liver cancer in New Jersey. Am J Epidemiol 117:443-454,1983 (20) AUSTIN H, DELZELL E, GRUFFERMAN S, ET AL:

A case-control study of hepatocellular carcinoma and the hepatitis B virus, cigarette smoking, and alcohol consumption. Cancer Res 46:962-966, 1986 (2/) FILIPPAZZO MG, ARAGONA E, COTTONE M, ET

(22) HARDELL L, BENGTSSON NO, JONSSON U, ET

AL: Aetiological aspects on primary liver cancer with special regard to alcohol, organic solvents, and acute intermittent porphyria: An epidemiological investigation. Br J Cancer 50:389-397,1984 (23)

TSUKUMA H, HlYAMA T, OSHIMA A , ET AL: A

case-control study of hepatocellular carcinoma in Osaka, Japan. Int J Cancer 45:231-236, 1990 (24) MAYANS MV, CALVET X, BRUIX J, ET AL: Risk

factors for hepatocellular carcinoma in Catalonia, Spain. Int J Cancer 46:378-381, 1990 (25) TRICHOPOULOS D, DAY NE, KAKLAMANI E, ET

AL: Hepatitis B virus, tobacco smoking, and ethanol consumption in the etiology of hepa-

Areca-derived A/-nitrosamines. Cancer Res 48:6912-6917,1988

(26) Lu SN, LJN TM, CHEN CJ, ET AL: A case-con-

(37) YAGER JD JR, YAGER R: Oral contraceptive

trol study of primary hepatocellular carcinoma in Taiwan. Cancer 62:2051-2055, 1988

steroids as promoters of hepatocarcinogenesis in female Sprague-Dawley rats. Cancer Res 40:3680-3685,1980 (38) Li JJ, LI SA: High incidence of hepatocellular carcinoma after synthetic estrogen administration in Syrian golden hamsters fed alpha-naphthoflavone: A new tumor model. JNCI 73:543-547,1984

(27) LA VECCHIA C, NEGRI E, DECARLI A, ET AL:

Risk factors for hepatocellular carcinoma in northern Italy. Int J Cancer 42:872-876, 1988 (28) KEW MC, DIBISCECLJE AM, PATERSON AC:

Smoking as a risk factor in hepatocellular carcinoma. A case-control study in southern African Blacks. Cancer 56:2315-2317, 1985 (29) HSING AW, MCLAUGHLIN JK, HRUBEC Z, ET

AL: Cigarette smoking and liver cancer among US veterans. Cancer Causes Control 1:217221, 1990 (50) MACMAHON B, PUGH TF: Epidemiology Prin-

ciples and Methods. Boston: Little, Brown, 1970, p 247 (31) DOLL R, HILL AB: A study of the etiology of

carcinoma of the lung. Br MedJ 2:1271-1286, 1952 (32) LnjENFELD AM, LIUENFELD DE: Foundations of Epidemiology. Oxford, England: Oxford Univ. Press, 1980, p 207 (33) ZIMMERMAN HJ: Hepatotoxicity: The Adverse Effects of Drugs and Other Chemicals on the Liver. New York: Appleton-Century-Crofts, 1978 (34) KlSELEVA NS, MlLIEVSKAJA IL, CAKLJN AV: Development of tumours in Syrian hamsters during prolonged experimental exposure to nas. Bull World Health Organ 54:597-605, 1976 (35) HECHT SS, CHEN CB, OHMORI T, ET AL: Com-

parative carcinogenicity in F344 rats of the tobacco-specific nitrosamines, AT-nitrosonornicotine and 4-{/V-methyl-A/-nitrosamino)-l-(3pyridyl)-l-butanone. Cancer Res 40:298-302, 1980 (36) RIVENSON A, HOFFMANN D, PROKOPCZYK B, ET

AL: Induction of lung and exocrine pancreas tumors in F344 rats by tobacco-specific and

(39) HENDERSON

BE,

PRESTON-MARTIN S, ED-

MONDSON HA, ET AL: Hepatocellular carcinoma and oral contraceptives. Br J Cancer 48:437-440,1983 (40) NEUBERGER J, FORMAN D, DOLL R, ET AL: Oral

contraceptives and hepatocellular carcinoma. BrMedJ[Clin Res] 292:1355-1357, 1986 (41) FORMAN D, VINCENT TJ, DOLL R: Cancer of

the liver and the use of oral contraceptives. Br MedJ[Clin Res] 292:1357-1361, 1986 (42) PALMER JR, ROSENBERG L, KAUFMAN DW, ET

AL: Oral contraceptive use and liver cancer. Am J Epidemiol 130:878-382, 1989 (43) LA VECCHIA C, NECRI E, PARAZZINI F: Oral

contraceptives and primary liver cancer. Br J Cancer 59:460-461, 1989 (44) THE WHO COLLABORATIVE STUDY OF NEOPLASIA AND STEROID CONTRACEPTIVES: Com-

bined oral contraceptives and liver cancer, hit J Cancer 43:254-259, 1989 (45) KEW MC, SONG E, MOHAMMED A, ET AL: Con-

traceptive steroids as a risk factor for hepatocellular carcinoma: A case/control study in South African black women. Hepatology 11:298-302,1990 (46) LA VECCHIA C, NEGRI E, D'AVANZO B, ET AL:

Medical history and primary liver cancer. Cancer Res 50:6274-6277, 1990 (47) LAWSON DH, GRAY JMB, MCKILLOPC, ETAL:

Diabetes mellitus and primary hepatocellular carcinoma. QJMed 61:945-955, 1986

Make Tracks

... to your nearest mailbox and send for the latest copy of the free Consumer Information Catalog.

It lists about 200 free or low-cost government publications on topics like health, nutrition, careers, money management, and federal benefits. Take a step in the right direction and write today for the free Consumer Information Catalog. Just send your name and address to: Consumer Information Center Department MT Pueblo, Colorado 81009

A public Mrvtce d life pabOcatioa and Uie Consumer Information Center of me US. General Services AdmlnblraUoa.

1826

Journal of the National Cancer Institute

Downloaded from http://jnci.oxfordjournals.org/ at Freie Universitaet Berlin on May 7, 2015

AL: Assessment of some risk factors for hepatocellular carcinoma: A case control study. StatMed 4:345-351, 1985

tocellular carcinoma. Int J Cancer 39:45-49, 1987

Nonviral risk factors for hepatocellular carcinoma in a low-risk population, the non-Asians of Los Angeles County, California.

We conducted interviews on 74 patients with histologically confirmed hepatocellular carcinoma. These patients, aged 18-74 years, were black or white r...
711KB Sizes 0 Downloads 0 Views