- Email: [email protected]
Primary liver cancer mortality in the United States
J ChronDis Vol. 36, No. 3, pp. 227-236, 1983 Printed
in Great
0021.968 I 83:030227-101603 0010 CopyrIght 0 1983 Pergamon Press Ltd
Britain. All rights reserved
PRIMARY
LIVER
DALE P. SANDLER’,
CANCER MORTALITY UNITED STATES
ROBERT S. SANDLER’
IN THE
and LOUISE F. HORNEY’
‘Epidemiology Branch, Biometry and Risk Assessment Program, National Institute of Environmental Health Sciences, P.O. Box 12233. Research Triangle Park. NC 27709 and ‘Division of Digestive Diseases, Department of Medicine. University of North Carolina School of Medicine, 324 Clinical Sciences Bldg 2298, Chapel Hill, NC 27514, U.S.A. (Received
in revised
form
27 Septmher
1982)
Abstract-We analyzed time trends and regional differences in mortality from primary liver cancer in the United States in order to explore possible effects of several factors which have been suggested as causes for liver cancer. Age-adjusted liver cancer death rates have been stable over time except for nonwhite males among whom there has been a 45% increase in liver cancer mortality between 1958 and 1975. The rates for nonwhite males are twice those for white males, and geographic trends in liver cancer mortality differ by race. The rates for white males are greatest in two southern regions. whereas rates for nonwhite males in these same regions are lower than they are elsewhere. Cirrhosis mortality trends parallel those for liver cancer among nonwhite males but not among white males. The similarities in trends for cirrhosis and liver cancer mortality among nonwhite males suggest that cirrhosis may be a major risk factor in this group. For white males, we must look to some other factor to explain the geographic differences that were observed.
INTRODUCTION
hepatocellular carcinoma (HCC) is an uncommon cancer in the United States, but is among the more common malignancies worldwide. Epidemiologic study of HCC in the United States has been difficult because of its rarity and because of problems of misclassification. Various reports have described liver cancer incidence and mortality in the United States as increasing or decreasing [l-4]. The inconsistency in these reports may be due to the fact that some of these reports failed to assess age-race-specific trends or surveyed selected areas of the country. A number of exposures have been assigned etiologic roles in HCC incidence, especially hepatitis B infection [S-S]. Alcohol, cirrhosis, aflatoxins and possibly organochlorine pesticides have also been suggested as causes [9-161. We analyzed time trends and regional differences in primary liver cancer mortality in the United States by age, race, and sex to assess the possible effects of some of these exposures. Trends in cirrhosis mortality and hepatitis B infection incidence were also explored in an attempt to explain the racial and regional differences in liver cancer mortality that we observed. PRIMARY
MATERIALS
AND
METHODS
Deaths from malignant neoplasms of the liver and intrahepatic bile ducts (ICD 155.0, 7th revision [17] and ICDA 155, 8th revision [lS] were obtained from United States Vital Statistics publications [19] for the years 1958 through 1975. Only approximately 5% of these deaths are attributed to neoplasms of intrahepatic bile ducts. No adjustment was made for differences that might result from changes in coding between the 7th and
Address correspondence to Dr Dale P. Sandler. This work was supported in part by a Public Health Service Preventive (Number CA00722) from the National Cancer Institute, DHHS. 227
Oncology
Academic
Award
Award
DALE
228
P.
SAWL~K
et t/l.
8th revisions, since the comparability ratio for codes 155 in the 8th revision and 155.0 in the 7th revision is reported to be 0.98 [20]. Death rates per 100,000 were calculated using census population figures for 1960 and 1970 and intercensal estimates published by the United States Census Bureau for non-census years [21.22]. Age-race-sex-specific liver cancer deaths by year and state were obtained from the National Center for Health Statistics (unpublished data) for the years 196X-1976. Death rates were calculated using 1970 age-raceesex-specific population data for each state. For regional comparisons, data were grouped into 3-yr periods, because the number of cases was too small to provide stable annual rates. Deaths attributed to liver cirrhosis (ICD 58 1, 7th revision and ICDA 571. 8th revision) for the time period 1958-1976 were obtained from United States Vital Statistics reports. Age-adjusted rates were calculated using direct adjustment. The 1970 combined United States population was used as the standard. Reported cases of hepatitis for each region were obtained from published reports 1231. Age-adjusted rates for 1975. which were not race or sex specific, were calculated. Reported race-specific cases were obtained for 1976 through 1979 from the National Hepatitis Surveillance Program (unpublished data), but were not available by age or XX. Time trends for hepatitis infection were not examined because of possible changes in reporting patterns over time. RESULTS
Death rates from liver cancer have changed little over the past 20 years, except for nonwhite males. The age-adjusted death rate for white males in the United States was 1.3/100,000 in 1958 and 1.5 in 1975. For nonwhite males, however, the age-adjusted death rate has risen from 2.5 to 3.6 over the same time period (Fig. 1). Although small in absolute terms, this represents an increase of 457: among nonwhite males which is statistically significant at the 5”/ level. The rates for nonwhite males were at least twice those for white males, but rates for females did not differ substantially by race. Death rates were higher among males than females: for nonwhites, rates for males were 3--4 times those for females, while for whites, the male risk was approximately twice that for females.
+
White
moles
+
White
trmales
-m-
Nonwhite
males
- l-
Nonwhite
females
l-.
,... \.
/
01
1955
I 1960
I 1965
I I970
I I975
Year
FIG. I. Time trends in age-adjusted death rates from primary race and sex, 1958-1975. (Direct adjustment, 1970 combined standard).
liver cancer in the United States by United States population used as the
Liver Cancer TABLE
In the U.S.
229
1. AGE-SPECIFIC LIVER CANCER DEATH KATES HY RACE AN" SEX PER ~00,000 POPULATION
White males 5G59 6G69 70-79 Nonwhite 50~_59 60-69 7&79
I960
1965
1970
1975
2.47 5.54 9.07
2.39 6.05 9.70
2.4 I 6.16 9.08
2.36 5.99 9.35
7.08 8.97 11.07
8.00 12.93 14.56
8.04 15.12 14.24
8.56 13.78 20.50
I .04 I .90
1.02 2.28 4.24
1.08 2.02 4.04
1.54 1.97 6.53
I .23 3.38 6.56
males
White females 50-59 6C-69 7&79
4.67
1.20 2.47 4.28
Nonwhite 50-59 60-69 7&79
I .27 3.29 3.05
2.1 I 3.48 4.04
females
Age-race-sex-specific death rates are given in Table 1 for four years; 1960, 1965, 1970 and 1975. Age-specific death rates for nonwhite males are greater than for white males in all time periods. The most dramatic increase over time has been for 6c-69 and 7&79 yr old nonwhite males, although the death rate for 70-79 yr old nonwhite females has also doubled from 1960 to 1975.
The states that comprise the various regions are given in Table 2. Regional differences in age-adjusted liver cancer mortality are shown in Fig. 2. For the combined United States population, the 3-yr age-adjusted primary liver cancer mortality rate was 3.7’100,OOO for 1974-1976. However, in the East South Central (R6) and West South Central (R7) regions. the corresponding 3-yr rates were 4.3 and 4.6. This difference between regional and United States rates is significant at the 5% level for the West South Central region and is of borderline significance for the East South Central region. Regional patterns differ strikingly by race (Fig. 3). Whereas rates for white males tend to be higher in the East South Central and West South Central regions, rates for nonwhites in these regions appear lower than in other regions. It is also interesting that while rates for white males in the East South Central and West South Central regions have increased from 1968-70 to 1974-76, liver cancer mortality for nonwhite males has decreased in those regions. Cirrhosis
rrrld hepatitis
Age-specific death rates from liver cirrhosis, shown in Fig. 4, follow a pattern similar to that for liver cancer mortality, although a racial difference in cirrhosis mortality is also prominent for females. Rates for nonwhites are approximately 1.5 times those for whites. Cirrhosis mortality for whites has increased only slightly over time, and appears to have leveled off. Nonwhite cirrhosis mortality has increased more dramatically, with the steepest increase in rates for nonwhite males. Again, the increasing trend appears to have tapered off. The age-adjusted death rate for nonwhite males rose from 13.5/100,000 in 1958 to a maximum of 34.3 in 1972. Death rates from cirrhosis appear to be lower in the South Central (R6, R7) regions than elsewhere (Fig. 5). This is similar to the pattern that was seen for liver cancer mortality among nonwhite males. Cirrhosis mortality for white males does not follow a geographic distribution similar to that for liver cancer mortality. The geographic distribution of reported hepatitis incidence does not correspond to the distribution of liver cancer deaths (Fig. 6). However, the incidence of reported hepatitis B among nonwhites (8.7/100,000 in 1976) is three times that for whites (2.9 in 1976).
230
DALE P. SANDLER et al. TABLI: 2. REGI~X OF THE UNITED STATES East South Central Alabama Kentucky Mississippi Tennessee
New England (RI) Connecticut Maine Massachusetts New Hampshire Rhode Island Vermont Middle Atlantic New Jersey New York Pennsylvania
West South Central Arkansas Louisiana Oklahoma Texas
(R2)
East North Central Illinois Indiana Michigan Ohio Wisconsin West North Central Iowa Kansas Minnesota Missour] Nebraska North Dakota South Dakota
(R3)
(R4)
(R6)
(R7)
Mountain (R8) Arizona Colorado Idaho Montana Nevada New Mexico Utah Wyoming Pacific (R9) Alaska California Hawaii Oregon Washington
South Atlantic (RS) Delaware District of Columbia Florida Georgia Maryland North Carolina South Carolina Virginia West Virginia
DISCUSSION
Mortality from HCC is probably underestimated using only deaths coded to ICDA 155.0, primary liver cancer. Studies have demonstrated that only approximately 50% of HCC deaths are coded as such 124,251. Percy [25] has shown that by including cancers coded after 1968 to ICDA 197.8, liver cancer not otherwise specified, the detection rate of
“r m
RI
R2
1968-1970
R3
R4
R5 R6 Region
R7
R6
R9
US
FIG. 2. Age-adjusted 3 yr primary hver cancer death rates by region of the United States and total United States in two time periods. (Direct adjustment, 1970 combined United States population used as the standard. Regions of the United States are given in Table 2).
Liver Cancer
‘31
in the U.S.
20
20 r
White
1st
males
White
18
females
F
m
1968 -
1970
1974 - 1976
3
16
_ 8
,2
~&I lo
m
1968-1970
m
1974-1976
i
4
f
6
P 0
4
2
2
0
0 RI
R2
R3
R4
R5
R6
R7
R8
R9
RI
US
R2
R3
R4
Nonwhite
R5
R6
RI
R8
R9
US
Region
Region
males
20 Nonwhite
females
I8
e 2
8
f
6
P 0
4 2
2
0
0 RI
R2
R3
R4
RS
R6
R7
R8
R9
US
RI
R2
R3
R4
Region
R5
R6
R7
R8
Region
FIG. 3. Age-adjusted 3 yr primary liver cancer death rates by region of the United States and total Unlted States in two time periods by race and sex. (Direct adjustment, 1970 combined United States population used as the standard. Regions of the United States are given in Table 2).
4035 ;; 8
_
+
White
males
-
White
females
-m-*-
Nonwhite Nonwhite
. .
males females
/
30I
c
.
n-a.
.
. ..
1970
1975
. ..9’
9.. g
25-
k ,”
20-
0, 5
l5-
/
c 6 :
/ IO6 5-
1955
1960
1965 Year
FIG.
4. Time
(Direct
in age-adjusted death rates from liver cirrhosis adjustment, 1970 combined United States population
trends
by race and sex. 1958-1975 used as the standard).
R9
US
232
DALE
P.
SANIILEK
or u/.
60
8 8
50
6 0
40
Nonwhite
t CL
males
30
al ‘0 L !z z
20
E IO
0 RI
R2
R3
R4
R5
R6
R7
R6
R9
Region FIG. 5. Age-adjusted liver cirrhosis death rates for males by region of the United States and race. 1970. (Direct adjustment, 1970 combined United States population used as the standard).
known primary liver cancer could be increased from 50 to 82”/). However. only 39”,,, of deaths coded to ICDA 197.8 could be confirmed as primary, whereas 735’,, of those coded to ICDA 155.0 could be confirmed as primary. For examining possible clues to etiology, it seems preferable to evaluate trends for ICDA 155.0 only. Although the absolute numbers are reduced substantially, the false positive rate is minimized. The accuracy of diagnosing HCC has improved due to the use of the z-fetoprotcin and newly developed diagnostic procedures. These advancements may be accompanied by ;I corresponding improvement in the accuracy of death certification. It is unclear what effect this would have on any apparent trends in liver cancer mortality prior to 1975. For the present. however, it is important to recognize that any trends suggested by vital statistics information must be validated by data from other sources. Our data show that liver cancer mortality for nonwhite males is about twice that for white males, and that nonwhite male death rates have been increasing. The discrepanq in liver cancer death rates for white males and nonwhite males suggests the existence of some differential exposure or susceptibility. Moreover. rates have increased only for nonwhite males suggesting that such a factor has become increasingly common among nonwhite males. Rates for nonwhite and white females do not differ substantially. suggesting that a genetic difference in risk may not be an important factor. WC also found that liver cancer death rates for whites were higher in the South Central (R6. R7) regions. However, for nonwhite males, the converse is true. 25
-0 w
3 0 0
20
m
Hepatitis
A
Hepatitis
B
Hepatitis
unsp ecifi
-
8 IS ; ,o =
IO
k ; I 5
5
0 RI
R2
R3
R4
R5
R6
R7
R9
R9
Region FIG. 6. Age-adjusted reported hepatitis incidence by region of the United States. 1975. (Direct adjustment. 1970 combined Unlted States population used as the standard).
Liver Cancer
in the U.S.
233
The rise in rates for nonwhites might be explained by improvements in reporting or medical care for nonwhites in recent years. Mortality from cancers of the esophagus and prostate have similarly shown no change over time for whites but have shown an increase for nonwhites. However, Devesa [26] found that these trends could not be explained by racial differences in accuracy of diagnosis, reporting or population enumeration. The lower rates among nonwhites in the South Central regions (where white rates are higher) might be due to inadequate reporting for nonwhites, but the further decline of death rates with time among nonwhite males in these regions is not plausibly explained by worsening records. In 1969, an additional category of liver cancer deaths, ICDA 197.8, was introduced to classify patients for whom liver cancer is not specified as primary or secondary. Any change in the number coded to ICDA 197.8 could affect the number coded to ICDA 155.0, primary liver cancer. It might be helpful in understanding trends in liver cancer mortality to also have data on trends in deaths coded to 197.8. It is more likely, however, that such a change would affect 197.7, liver cancer specified as secondary. Furthermore, if racial or geographical differences in liver cancer mortality are due to changes or regional differences in the use of 197.8, changes in liver cancer mortality would most likely be consistent within regions or races. However, the data here show that liver cancer mortality has increased for nonwhite males only and racial patterns are not consistent within regions. An etiologic relationship of liver cirrhosis to HCC has long been suspected because the two conditions often occur together [3]. In different reports, 50-90% of patients with HCC have cirrhosis [9. 10, 16.271. Conversely, from 5 to 40% of cirrhotics are reported to develop HCC [ 10,271. Because of the association of HCC with cirrhosis, we explored trends in cirrhosis mortality to attempt to understand the trends in liver cancer mortality that we observed. Mortality from liver cirrhosis has increased steadily through 1972, with the most dramatic increase for nonwhite males. Our data for liver cancer mortality show similar trends for nonwhites, although the increase in liver cancer mortality has not been as great as that for cirrhosis mortality. Cirrhosis mortality has increased significantly for nonwhite females, but there has been no apparent corresponding increase in liver cancer mortality (although age adjusted rates after 1965 are about 20% higher than rates before 1965). Liver cancer death rates for nonwhite females are based on small numbers and the instability of these rates makes it difficult to assess time trends. For whites, cirrhosis, but not liver cancer, mortality has shown a slightly increasing trend. Liver cancer mortality for nonwhite males is twice that for white males. The race-ratio for cirrhosis mortality is 1.5 and that for cirrhosis incidence is reported to be 1.75 [28]. Thus, both time trends and race-ratios for cirrhosis and liver cancer mortality support an association between the two diseases for nonwhite males. In our regional comparisons, cirrhosis mortality was lower in the South Central (R6, R7) regions for nonwhite males, a trend echoed by liver cancer mortality. For white males, geographic patterns for liver cancer and cirrhosis mortality are not parallel. Cirrhosis may be an important factor in liver cancer mortality for nonwhites, but not for whites. Garagliano et al. [28] found that peak cirrhosis incidence was 20 yr earlier for nonwhite males (age 40-49 vs age 60-69) than for white males. United States Vital Statistics data also show that peak cirrhosis mortality occurs earlier among nonwhite males. If the latent interval between cirrhosis and HCC is long, competing causes of death for the older whites with cirrhosis would decrease the chance of developing HCC. Race or sex differences in disease severity or prognosis with cirrhosis might also influence HCC mortality statistics. Persons with more fulminant cirrhosis will be less likely to survive long enough to develop HCC. Competing mortality, then, might explain the lack of parallel cirrhosis and liver cancer mortality for whites and nonwhite females. In the United States, cirrhosis is more often associated with alcohol consumption than with hepatitis 191. Trends in alcohol consumption between 1961 and 1971 are parallel to the trend seen for cirrhosis [29]. When average age adjusted cancer mortality rates for
2.74
DALE P. SANDLEK rt d.
19.50P1967 were correlated with per capita consumption of spirits, wine, and beer as estimated from tax receipts in 1960, however, liver cancer was only in excess among women who drank beer [30]. Racial differences in liver cancer mortality are not easily explained by differences in alcohol consumption. Nonwhites have been variously described as drinking more [3 l] or less [29,32,33] than whites. The issue of the role played by alcohol in the etiology of liver cancer is complicated by the potential involvement of hepatitis B virus. Recent evidence suggests that hepatitis B virus may play an important role in the pathogenesis of liver cancer in alcoholics [34]. A great deal of evidence has been gathered supporting hepatitis B virus as a cause for HCC. In the United States, chronic hepatitis antigen is found in less than l’;;, of individuals, although the attributable risk may still be high. The prevalence of hepatitis B virus among HCC patients has been demonstrated to be greater than among patients with other cancers or among normal controls [ 1 1, 351. Similarly, evidence of hepatitis B infection is found more commonly among patients with cirrhosis than among normal controls and the frequency of hepatitis B markers is similar among cirrhotics and patients with HCC [S, 63. Cirrhosis may. therefore, be an intermediate step in a causal link between hepatitis and cancer 161. It is difficult to assess data on trends in hepatitis B infection because the assay for hepatitis B antigen has only recently become widely available. There seems to be a trend in many areas toward an increased incidence of type B infection [7]. We found that there were no regional differences in hepatitis that might explain high rates of HCC in the South Central (R6, R7) regions. Reported hepatitis B virus incidence rates for nonwhites are, however, three times those for whites, corresponding to the greater liver cancer mortality among nonwhites. It is important to recognize that viral hepatitis may be significantly under-reported for both hospitalized and nonhospitalized patients, and racial differences could be attributed to differences in reporting. Aflatoxins are toxins produced by an Aspergillus fungus which grows on plants such as oats, soybeans, corn, rye, cotton and peanuts. Aflatoxins are the most potent hepatocarcinogen per unit weight in many species 121. They were initially suspected as agents for human HCC when they were found to contaminate foodstuffs in countries where HCC was particularly common. Quantitative studies from Thailand, Uganda, Switzerland. Kenya and Mozambique have shown a doseeresponse relationship between consumption of contaminated food and HCC rates. with greatest danger in hot, moist climates [ 12,36-381. Urinary aflatoxin levels have been found to be 40 times higher than normal in patients with hepatitis, cirrhosis and HCC [39]. Contamination with food spoilage fungi is more likely in underdeveloped countries. In the United States, chronic exposure to aflatoxins is most likely to have occurred in rural areas of the South where unmarketable peanuts and corn are used for local consumption. FDA surveys have shown that aflatoxin contamination of corn is a problem primarily in the southeastern states, that most of the contamination is of field origin, and that exposure has probably been endemic in that area [ 131. One might speculate that the higher rates of liver cancer found in the South Central regions are due to aflatoxins. The liver cancer excess in the South Central regions was observed only for whites. For nonwhites, cirrhosis seems to be an overriding factor in liver cancer mortality and an association with other factors may be obscured. The South Atlantic (R5) region includes both predominantly agricultural and predominantly industrial states, and this heterogeneity may explain why rates in this region were not also elevated. Another feature which might distinguish the South would be pesticide exposure. While there has been no convincing evidence that these agents induce tumors in humans. they have been found to be hepatocarcinogens in animals [14]. Regional differences in organochlorine pesticide use might explain regional differences in liver cancer mortality, but racial differences within regions might be difficult to explain on the basis of pesticide exposure. Factors that have not been addressed, such as the increasing industralization of the Southern regions and different employment patterns for whites and nonwhites might also
Liver Cancer
in the U.S.
235
influence HCC mortality trends. Thus, it is not possible to make definite conclusions from the data presented here. The apparent similarities in trends for cirrhosis and liver cancer mortality among nonwhite males suggests that cirrhosis may be a major risk factor for this group. For white males, among whom cirrhosis mortality is less common than among nonwhites and occurs at a later age, we must look to some other factor to explain the geographic differences observed. Such a factor might also play an etiologic role in the development of liver cancer in nonwhite males, but may be difficult to detect because of the apparently strong relationship to cirrhosis.
REFERENCES I.
2. 3. 4. 5. 6. 7. 8.
9. IO.
I I. 17. 13. 14. 15. 16. 17.
IX. 19. 70. 71. 22. 23. 24. 25. 26. 27.
28. 29.
Devesa SS. Silverman DT: Cancer incidence and mortality trends in the United States: 193551974. J Nat1 Cancer lnst60: 5455571, 1978 Saracci R. Repetto F: Time trends of primary liver cancer: indication of increased incidence in selected cancer registry populations. J Nat1 Cancer lnst 65: 241-247, 1980 Deichmann WB, MacDonald WE: Organochlorine pesticides and liver cancer deaths in the United States, 1930~-1972. Ecotoxic. Envir Safety I :89-l IO, 1977 Aoki K: Cancer of the liver, international mortality trends. World Hlth Stat Rep 3 I : 28-50, 1978 Blumberg BS. London WT: Hepatitis B virus and the prevention of primary hepatocellular carcinoma. N Engl J Med 304, 782-784. 1981 Szmuness W: Hepatocellular carcinoma and the hepatitis B virus: Evidence for a causal association. Prog Med Virol 24: 4&69, 1978 Zuckerman AJ: Hepatitis B virus and primary liver cancer. In Primary Liver Tumors. Remmer H, Bolt HM. Bannasch P, Popper H (Eds), Falk Symposium. Baltimore: University Park Press. pp. 4255430, 1978 Okuda K, Nakashima T: Hepatocellular carcinoma: A review of the recent studies and developments. In Progress in Liver Diseases. Popper H, Schaffner F (Eds). New York: Grune & Stratton, Vol. VI. pp. 6399650. I979 Shikata T: Primary liver carcinoma and liver cirrhosis. In Hepatocellular Carcinoma. Okuda K, Peters RL (Eds). New York: John Wiley. pp. 53-71. 1976 Hutt MSR: Epidemiology of human primary liver cancer. In International Agency for Research on Cancer. Liver Cancer IARC Sci. Pub. No. I, pp. 21-29, Lyon 1971 Omata M, Ashcaval M. Liew C: Hepatocellular carcinoma in the USA: Etiologic considerations. Gastroenterology 76: 279-287, 1979 Popper H: Hepatic cancers in man: Quantitative perspectives. Envir Res 19: 4822494. 1979 Stoloff L. Friedman L: Information bearing on the evaluation 01 the hazard to man from atlatoxin ingestion. PAG Bull 6: 21 37. 1976 Wogan GN: Aflatoxins and their relationship to hepatocellular carcinoma. In Hepatocellular Carcinoma Okuda K. Peters RL (Eds). New York: John Wilev, PP. 25542. 1976 Shank RC: Epidemiology of aflatoxin carcinogenksis.~ In Advances in Modern Toxicology Kraybill HF. Mehlman MA (Eds). New York: John Wiley. Vol. 3, pp. 291-318, 1977 Scheuer PJ : Pathopenesis and morbid anatomy of human primary liver cancer. In International Agency for Research on Cnncer. Liver Cancer. IARC Scientific Publication No. 1, pp. 13-20, Lyon, 1971 _ World Health Organization: Manual of the International Statistical Classification of Diseases, Iniuries. and Causes of Death,Vol. I. Based on the recommendations of the Seventh Revision Conference, 1955. Geneva, Switzerland, 1957 National Center for Health Statistics: International Classification of Diseases, adapted for use in the United States. DHEW, PHS Pub. No. 1693, 1968 National Center for Health Statistics: Vital Statistics of the United States: Mortality, 1958-1975. DHEW, PHS, Rockville. MD. Kleba AJ. Dolman AB: Comparability of mortality statistics for the seventh and eighth revision of the international classification of diseases. DHEW Pub. No. HRA/76-1340, Vital Statistics Series No. 2, 1975 U.S. Department of Commerce, Bureau of Census: 1970 Census of Population, Vol. 1, Characteristics of the population. Part 1. United States Summary. Section I. 1973 US Department of Commerce. Bureau of Census: Current Population Reports. Series P-25, Estimates of the population of the United States by age, race and sex. 1958-1975 US Center for Disease Control: Morbidity and Mortality Weekly Report, annual supplement. summary 1975. Vol. 24. No. 54. US DHEW No. (CDC) 76-8241, 1976 Hammond EC. Johnson H, Percy C: Reporting of cancer of the lung, liver and uterus. Cancer IO: 1802~1806, 1968 Percy C. Stanek E, Gloeckler L: Accuracy of cancer death certificates and its effect on cancer mortahty statistics. Am J Publ Hltb 71: 2422250. 1981 Devesa SS: Reliability of reported death rates and incidence rates. Prev Med 9: 5899600. 1980 Lehmann FG, Wegencr T: Etiology of human liver cancer: Controlled prospective study in liver cirrhosis. In Liver Carcinogenesis Lapis K, Johannessen JV (Eds). Washington: Hemisphere Publishing Corp, pp. I1 1~~129. 1979 Gurngliano CF. Lilienfeld AM, Mendeloff AI: Incidence rates of liver cirrhosis and related diseases in Baltimore and selected areas of the United States. J Chron Dis 32: 5433554. 1979 Noble EP: Third Special Report to the Congress on Alcohol and Health. Rockville, Maryland. US DHEW, National lnstitute on Alcohol Abuse and Alcoholism. 1978
DALE P.SANDLERettrl.
236
30. Breslow NE, Enstrom 31. 32. 33.
34.
35. 36. 37. 38. 39.
JE: Geographic correlations between cancer mortality rates and alcoho~tobacco consumption in the United States. J Nat1 Cancer Inst 53: 631-639, 1974 Keller A: Alcoholic beverages and carcinogenesis. NY State J Med 80: 1245-1250, 19X0 Mettlin C: Nutritional habits of blacks and whites. Prev Med 9: 601 -606, 1980 Klatsky AL. Friedman GD, Siegelaub AB, Gerard MJ: Alcohol consumption among white. black or oriental men and women: Kaiser-Permanente multiphnsic health examination data. Am J Epid 105: 31 l-323, 1977 Brechot C. Nalpus B, Courouce A, Duhamel G. Callard P. Carnot F, Tiollais P. Berthelot P: Evidence that hepatitis B virus has a role in liver-cell carcmoma in alcoholic liver disease. N Engl J Med 306: 1384-1387, 1982 Yarrish RL, Werner BG, Blumberg BS: Association of Hepatitis B virus infection with hepatocellular carcinoma in American patients. Int J Cancer 26: 71 I-715. 1980 Keen P. Martin P: Is aflatoxin carcinogenic in man’? The evidence in Swazil;lnd. Trap Geogr Med 23: 49-53. 1971 Alpert ME. Hutt MSR, Wogan GN: Association between aflntoxin contamination of food and hepatoma frequency in Uganda. Cancer 28: 253-260, 1971 Peers FG, Linsell CA: Dietary aflatoxins and liver cancer: A population based study in Kenya. Br J Cancer 27: 473-489, 1973 Bababunmi EA: Excretion of aflatoxin in the urine of normal individuals and patients with liver diseases in Ibudan (Nigeria). In International Study Group for the Detection and Prevention of Cancer. Prevention and Detection of cancer. Proc 3rd Int Symp Detection and Prevention of Cancer New York. Vol. 2. Part I, pp. 1729-1736, 1978
in Great
0021.968 I 83:030227-101603 0010 CopyrIght 0 1983 Pergamon Press Ltd
Britain. All rights reserved
PRIMARY
LIVER
DALE P. SANDLER’,
CANCER MORTALITY UNITED STATES
ROBERT S. SANDLER’
IN THE
and LOUISE F. HORNEY’
‘Epidemiology Branch, Biometry and Risk Assessment Program, National Institute of Environmental Health Sciences, P.O. Box 12233. Research Triangle Park. NC 27709 and ‘Division of Digestive Diseases, Department of Medicine. University of North Carolina School of Medicine, 324 Clinical Sciences Bldg 2298, Chapel Hill, NC 27514, U.S.A. (Received
in revised
form
27 Septmher
1982)
Abstract-We analyzed time trends and regional differences in mortality from primary liver cancer in the United States in order to explore possible effects of several factors which have been suggested as causes for liver cancer. Age-adjusted liver cancer death rates have been stable over time except for nonwhite males among whom there has been a 45% increase in liver cancer mortality between 1958 and 1975. The rates for nonwhite males are twice those for white males, and geographic trends in liver cancer mortality differ by race. The rates for white males are greatest in two southern regions. whereas rates for nonwhite males in these same regions are lower than they are elsewhere. Cirrhosis mortality trends parallel those for liver cancer among nonwhite males but not among white males. The similarities in trends for cirrhosis and liver cancer mortality among nonwhite males suggest that cirrhosis may be a major risk factor in this group. For white males, we must look to some other factor to explain the geographic differences that were observed.
INTRODUCTION
hepatocellular carcinoma (HCC) is an uncommon cancer in the United States, but is among the more common malignancies worldwide. Epidemiologic study of HCC in the United States has been difficult because of its rarity and because of problems of misclassification. Various reports have described liver cancer incidence and mortality in the United States as increasing or decreasing [l-4]. The inconsistency in these reports may be due to the fact that some of these reports failed to assess age-race-specific trends or surveyed selected areas of the country. A number of exposures have been assigned etiologic roles in HCC incidence, especially hepatitis B infection [S-S]. Alcohol, cirrhosis, aflatoxins and possibly organochlorine pesticides have also been suggested as causes [9-161. We analyzed time trends and regional differences in primary liver cancer mortality in the United States by age, race, and sex to assess the possible effects of some of these exposures. Trends in cirrhosis mortality and hepatitis B infection incidence were also explored in an attempt to explain the racial and regional differences in liver cancer mortality that we observed. PRIMARY
MATERIALS
AND
METHODS
Deaths from malignant neoplasms of the liver and intrahepatic bile ducts (ICD 155.0, 7th revision [17] and ICDA 155, 8th revision [lS] were obtained from United States Vital Statistics publications [19] for the years 1958 through 1975. Only approximately 5% of these deaths are attributed to neoplasms of intrahepatic bile ducts. No adjustment was made for differences that might result from changes in coding between the 7th and
Address correspondence to Dr Dale P. Sandler. This work was supported in part by a Public Health Service Preventive (Number CA00722) from the National Cancer Institute, DHHS. 227
Oncology
Academic
Award
Award
DALE
228
P.
SAWL~K
et t/l.
8th revisions, since the comparability ratio for codes 155 in the 8th revision and 155.0 in the 7th revision is reported to be 0.98 [20]. Death rates per 100,000 were calculated using census population figures for 1960 and 1970 and intercensal estimates published by the United States Census Bureau for non-census years [21.22]. Age-race-sex-specific liver cancer deaths by year and state were obtained from the National Center for Health Statistics (unpublished data) for the years 196X-1976. Death rates were calculated using 1970 age-raceesex-specific population data for each state. For regional comparisons, data were grouped into 3-yr periods, because the number of cases was too small to provide stable annual rates. Deaths attributed to liver cirrhosis (ICD 58 1, 7th revision and ICDA 571. 8th revision) for the time period 1958-1976 were obtained from United States Vital Statistics reports. Age-adjusted rates were calculated using direct adjustment. The 1970 combined United States population was used as the standard. Reported cases of hepatitis for each region were obtained from published reports 1231. Age-adjusted rates for 1975. which were not race or sex specific, were calculated. Reported race-specific cases were obtained for 1976 through 1979 from the National Hepatitis Surveillance Program (unpublished data), but were not available by age or XX. Time trends for hepatitis infection were not examined because of possible changes in reporting patterns over time. RESULTS
Death rates from liver cancer have changed little over the past 20 years, except for nonwhite males. The age-adjusted death rate for white males in the United States was 1.3/100,000 in 1958 and 1.5 in 1975. For nonwhite males, however, the age-adjusted death rate has risen from 2.5 to 3.6 over the same time period (Fig. 1). Although small in absolute terms, this represents an increase of 457: among nonwhite males which is statistically significant at the 5”/ level. The rates for nonwhite males were at least twice those for white males, but rates for females did not differ substantially by race. Death rates were higher among males than females: for nonwhites, rates for males were 3--4 times those for females, while for whites, the male risk was approximately twice that for females.
+
White
moles
+
White
trmales
-m-
Nonwhite
males
- l-
Nonwhite
females
l-.
,... \.
/
01
1955
I 1960
I 1965
I I970
I I975
Year
FIG. I. Time trends in age-adjusted death rates from primary race and sex, 1958-1975. (Direct adjustment, 1970 combined standard).
liver cancer in the United States by United States population used as the
Liver Cancer TABLE
In the U.S.
229
1. AGE-SPECIFIC LIVER CANCER DEATH KATES HY RACE AN" SEX PER ~00,000 POPULATION
White males 5G59 6G69 70-79 Nonwhite 50~_59 60-69 7&79
I960
1965
1970
1975
2.47 5.54 9.07
2.39 6.05 9.70
2.4 I 6.16 9.08
2.36 5.99 9.35
7.08 8.97 11.07
8.00 12.93 14.56
8.04 15.12 14.24
8.56 13.78 20.50
I .04 I .90
1.02 2.28 4.24
1.08 2.02 4.04
1.54 1.97 6.53
I .23 3.38 6.56
males
White females 50-59 6C-69 7&79
4.67
1.20 2.47 4.28
Nonwhite 50-59 60-69 7&79
I .27 3.29 3.05
2.1 I 3.48 4.04
females
Age-race-sex-specific death rates are given in Table 1 for four years; 1960, 1965, 1970 and 1975. Age-specific death rates for nonwhite males are greater than for white males in all time periods. The most dramatic increase over time has been for 6c-69 and 7&79 yr old nonwhite males, although the death rate for 70-79 yr old nonwhite females has also doubled from 1960 to 1975.
The states that comprise the various regions are given in Table 2. Regional differences in age-adjusted liver cancer mortality are shown in Fig. 2. For the combined United States population, the 3-yr age-adjusted primary liver cancer mortality rate was 3.7’100,OOO for 1974-1976. However, in the East South Central (R6) and West South Central (R7) regions. the corresponding 3-yr rates were 4.3 and 4.6. This difference between regional and United States rates is significant at the 5% level for the West South Central region and is of borderline significance for the East South Central region. Regional patterns differ strikingly by race (Fig. 3). Whereas rates for white males tend to be higher in the East South Central and West South Central regions, rates for nonwhites in these regions appear lower than in other regions. It is also interesting that while rates for white males in the East South Central and West South Central regions have increased from 1968-70 to 1974-76, liver cancer mortality for nonwhite males has decreased in those regions. Cirrhosis
rrrld hepatitis
Age-specific death rates from liver cirrhosis, shown in Fig. 4, follow a pattern similar to that for liver cancer mortality, although a racial difference in cirrhosis mortality is also prominent for females. Rates for nonwhites are approximately 1.5 times those for whites. Cirrhosis mortality for whites has increased only slightly over time, and appears to have leveled off. Nonwhite cirrhosis mortality has increased more dramatically, with the steepest increase in rates for nonwhite males. Again, the increasing trend appears to have tapered off. The age-adjusted death rate for nonwhite males rose from 13.5/100,000 in 1958 to a maximum of 34.3 in 1972. Death rates from cirrhosis appear to be lower in the South Central (R6, R7) regions than elsewhere (Fig. 5). This is similar to the pattern that was seen for liver cancer mortality among nonwhite males. Cirrhosis mortality for white males does not follow a geographic distribution similar to that for liver cancer mortality. The geographic distribution of reported hepatitis incidence does not correspond to the distribution of liver cancer deaths (Fig. 6). However, the incidence of reported hepatitis B among nonwhites (8.7/100,000 in 1976) is three times that for whites (2.9 in 1976).
230
DALE P. SANDLER et al. TABLI: 2. REGI~X OF THE UNITED STATES East South Central Alabama Kentucky Mississippi Tennessee
New England (RI) Connecticut Maine Massachusetts New Hampshire Rhode Island Vermont Middle Atlantic New Jersey New York Pennsylvania
West South Central Arkansas Louisiana Oklahoma Texas
(R2)
East North Central Illinois Indiana Michigan Ohio Wisconsin West North Central Iowa Kansas Minnesota Missour] Nebraska North Dakota South Dakota
(R3)
(R4)
(R6)
(R7)
Mountain (R8) Arizona Colorado Idaho Montana Nevada New Mexico Utah Wyoming Pacific (R9) Alaska California Hawaii Oregon Washington
South Atlantic (RS) Delaware District of Columbia Florida Georgia Maryland North Carolina South Carolina Virginia West Virginia
DISCUSSION
Mortality from HCC is probably underestimated using only deaths coded to ICDA 155.0, primary liver cancer. Studies have demonstrated that only approximately 50% of HCC deaths are coded as such 124,251. Percy [25] has shown that by including cancers coded after 1968 to ICDA 197.8, liver cancer not otherwise specified, the detection rate of
“r m
RI
R2
1968-1970
R3
R4
R5 R6 Region
R7
R6
R9
US
FIG. 2. Age-adjusted 3 yr primary hver cancer death rates by region of the United States and total United States in two time periods. (Direct adjustment, 1970 combined United States population used as the standard. Regions of the United States are given in Table 2).
Liver Cancer
‘31
in the U.S.
20
20 r
White
1st
males
White
18
females
F
m
1968 -
1970
1974 - 1976
3
16
_ 8
,2
~&I lo
m
1968-1970
m
1974-1976
i
4
f
6
P 0
4
2
2
0
0 RI
R2
R3
R4
R5
R6
R7
R8
R9
RI
US
R2
R3
R4
Nonwhite
R5
R6
RI
R8
R9
US
Region
Region
males
20 Nonwhite
females
I8
e 2
8
f
6
P 0
4 2
2
0
0 RI
R2
R3
R4
RS
R6
R7
R8
R9
US
RI
R2
R3
R4
Region
R5
R6
R7
R8
Region
FIG. 3. Age-adjusted 3 yr primary liver cancer death rates by region of the United States and total Unlted States in two time periods by race and sex. (Direct adjustment, 1970 combined United States population used as the standard. Regions of the United States are given in Table 2).
4035 ;; 8
_
+
White
males
-
White
females
-m-*-
Nonwhite Nonwhite
. .
males females
/
30I
c
.
n-a.
.
. ..
1970
1975
. ..9’
9.. g
25-
k ,”
20-
0, 5
l5-
/
c 6 :
/ IO6 5-
1955
1960
1965 Year
FIG.
4. Time
(Direct
in age-adjusted death rates from liver cirrhosis adjustment, 1970 combined United States population
trends
by race and sex. 1958-1975 used as the standard).
R9
US
232
DALE
P.
SANIILEK
or u/.
60
8 8
50
6 0
40
Nonwhite
t CL
males
30
al ‘0 L !z z
20
E IO
0 RI
R2
R3
R4
R5
R6
R7
R6
R9
Region FIG. 5. Age-adjusted liver cirrhosis death rates for males by region of the United States and race. 1970. (Direct adjustment, 1970 combined United States population used as the standard).
known primary liver cancer could be increased from 50 to 82”/). However. only 39”,,, of deaths coded to ICDA 197.8 could be confirmed as primary, whereas 735’,, of those coded to ICDA 155.0 could be confirmed as primary. For examining possible clues to etiology, it seems preferable to evaluate trends for ICDA 155.0 only. Although the absolute numbers are reduced substantially, the false positive rate is minimized. The accuracy of diagnosing HCC has improved due to the use of the z-fetoprotcin and newly developed diagnostic procedures. These advancements may be accompanied by ;I corresponding improvement in the accuracy of death certification. It is unclear what effect this would have on any apparent trends in liver cancer mortality prior to 1975. For the present. however, it is important to recognize that any trends suggested by vital statistics information must be validated by data from other sources. Our data show that liver cancer mortality for nonwhite males is about twice that for white males, and that nonwhite male death rates have been increasing. The discrepanq in liver cancer death rates for white males and nonwhite males suggests the existence of some differential exposure or susceptibility. Moreover. rates have increased only for nonwhite males suggesting that such a factor has become increasingly common among nonwhite males. Rates for nonwhite and white females do not differ substantially. suggesting that a genetic difference in risk may not be an important factor. WC also found that liver cancer death rates for whites were higher in the South Central (R6. R7) regions. However, for nonwhite males, the converse is true. 25
-0 w
3 0 0
20
m
Hepatitis
A
Hepatitis
B
Hepatitis
unsp ecifi
-
8 IS ; ,o =
IO
k ; I 5
5
0 RI
R2
R3
R4
R5
R6
R7
R9
R9
Region FIG. 6. Age-adjusted reported hepatitis incidence by region of the United States. 1975. (Direct adjustment. 1970 combined Unlted States population used as the standard).
Liver Cancer
in the U.S.
233
The rise in rates for nonwhites might be explained by improvements in reporting or medical care for nonwhites in recent years. Mortality from cancers of the esophagus and prostate have similarly shown no change over time for whites but have shown an increase for nonwhites. However, Devesa [26] found that these trends could not be explained by racial differences in accuracy of diagnosis, reporting or population enumeration. The lower rates among nonwhites in the South Central regions (where white rates are higher) might be due to inadequate reporting for nonwhites, but the further decline of death rates with time among nonwhite males in these regions is not plausibly explained by worsening records. In 1969, an additional category of liver cancer deaths, ICDA 197.8, was introduced to classify patients for whom liver cancer is not specified as primary or secondary. Any change in the number coded to ICDA 197.8 could affect the number coded to ICDA 155.0, primary liver cancer. It might be helpful in understanding trends in liver cancer mortality to also have data on trends in deaths coded to 197.8. It is more likely, however, that such a change would affect 197.7, liver cancer specified as secondary. Furthermore, if racial or geographical differences in liver cancer mortality are due to changes or regional differences in the use of 197.8, changes in liver cancer mortality would most likely be consistent within regions or races. However, the data here show that liver cancer mortality has increased for nonwhite males only and racial patterns are not consistent within regions. An etiologic relationship of liver cirrhosis to HCC has long been suspected because the two conditions often occur together [3]. In different reports, 50-90% of patients with HCC have cirrhosis [9. 10, 16.271. Conversely, from 5 to 40% of cirrhotics are reported to develop HCC [ 10,271. Because of the association of HCC with cirrhosis, we explored trends in cirrhosis mortality to attempt to understand the trends in liver cancer mortality that we observed. Mortality from liver cirrhosis has increased steadily through 1972, with the most dramatic increase for nonwhite males. Our data for liver cancer mortality show similar trends for nonwhites, although the increase in liver cancer mortality has not been as great as that for cirrhosis mortality. Cirrhosis mortality has increased significantly for nonwhite females, but there has been no apparent corresponding increase in liver cancer mortality (although age adjusted rates after 1965 are about 20% higher than rates before 1965). Liver cancer death rates for nonwhite females are based on small numbers and the instability of these rates makes it difficult to assess time trends. For whites, cirrhosis, but not liver cancer, mortality has shown a slightly increasing trend. Liver cancer mortality for nonwhite males is twice that for white males. The race-ratio for cirrhosis mortality is 1.5 and that for cirrhosis incidence is reported to be 1.75 [28]. Thus, both time trends and race-ratios for cirrhosis and liver cancer mortality support an association between the two diseases for nonwhite males. In our regional comparisons, cirrhosis mortality was lower in the South Central (R6, R7) regions for nonwhite males, a trend echoed by liver cancer mortality. For white males, geographic patterns for liver cancer and cirrhosis mortality are not parallel. Cirrhosis may be an important factor in liver cancer mortality for nonwhites, but not for whites. Garagliano et al. [28] found that peak cirrhosis incidence was 20 yr earlier for nonwhite males (age 40-49 vs age 60-69) than for white males. United States Vital Statistics data also show that peak cirrhosis mortality occurs earlier among nonwhite males. If the latent interval between cirrhosis and HCC is long, competing causes of death for the older whites with cirrhosis would decrease the chance of developing HCC. Race or sex differences in disease severity or prognosis with cirrhosis might also influence HCC mortality statistics. Persons with more fulminant cirrhosis will be less likely to survive long enough to develop HCC. Competing mortality, then, might explain the lack of parallel cirrhosis and liver cancer mortality for whites and nonwhite females. In the United States, cirrhosis is more often associated with alcohol consumption than with hepatitis 191. Trends in alcohol consumption between 1961 and 1971 are parallel to the trend seen for cirrhosis [29]. When average age adjusted cancer mortality rates for
2.74
DALE P. SANDLEK rt d.
19.50P1967 were correlated with per capita consumption of spirits, wine, and beer as estimated from tax receipts in 1960, however, liver cancer was only in excess among women who drank beer [30]. Racial differences in liver cancer mortality are not easily explained by differences in alcohol consumption. Nonwhites have been variously described as drinking more [3 l] or less [29,32,33] than whites. The issue of the role played by alcohol in the etiology of liver cancer is complicated by the potential involvement of hepatitis B virus. Recent evidence suggests that hepatitis B virus may play an important role in the pathogenesis of liver cancer in alcoholics [34]. A great deal of evidence has been gathered supporting hepatitis B virus as a cause for HCC. In the United States, chronic hepatitis antigen is found in less than l’;;, of individuals, although the attributable risk may still be high. The prevalence of hepatitis B virus among HCC patients has been demonstrated to be greater than among patients with other cancers or among normal controls [ 1 1, 351. Similarly, evidence of hepatitis B infection is found more commonly among patients with cirrhosis than among normal controls and the frequency of hepatitis B markers is similar among cirrhotics and patients with HCC [S, 63. Cirrhosis may. therefore, be an intermediate step in a causal link between hepatitis and cancer 161. It is difficult to assess data on trends in hepatitis B infection because the assay for hepatitis B antigen has only recently become widely available. There seems to be a trend in many areas toward an increased incidence of type B infection [7]. We found that there were no regional differences in hepatitis that might explain high rates of HCC in the South Central (R6, R7) regions. Reported hepatitis B virus incidence rates for nonwhites are, however, three times those for whites, corresponding to the greater liver cancer mortality among nonwhites. It is important to recognize that viral hepatitis may be significantly under-reported for both hospitalized and nonhospitalized patients, and racial differences could be attributed to differences in reporting. Aflatoxins are toxins produced by an Aspergillus fungus which grows on plants such as oats, soybeans, corn, rye, cotton and peanuts. Aflatoxins are the most potent hepatocarcinogen per unit weight in many species 121. They were initially suspected as agents for human HCC when they were found to contaminate foodstuffs in countries where HCC was particularly common. Quantitative studies from Thailand, Uganda, Switzerland. Kenya and Mozambique have shown a doseeresponse relationship between consumption of contaminated food and HCC rates. with greatest danger in hot, moist climates [ 12,36-381. Urinary aflatoxin levels have been found to be 40 times higher than normal in patients with hepatitis, cirrhosis and HCC [39]. Contamination with food spoilage fungi is more likely in underdeveloped countries. In the United States, chronic exposure to aflatoxins is most likely to have occurred in rural areas of the South where unmarketable peanuts and corn are used for local consumption. FDA surveys have shown that aflatoxin contamination of corn is a problem primarily in the southeastern states, that most of the contamination is of field origin, and that exposure has probably been endemic in that area [ 131. One might speculate that the higher rates of liver cancer found in the South Central regions are due to aflatoxins. The liver cancer excess in the South Central regions was observed only for whites. For nonwhites, cirrhosis seems to be an overriding factor in liver cancer mortality and an association with other factors may be obscured. The South Atlantic (R5) region includes both predominantly agricultural and predominantly industrial states, and this heterogeneity may explain why rates in this region were not also elevated. Another feature which might distinguish the South would be pesticide exposure. While there has been no convincing evidence that these agents induce tumors in humans. they have been found to be hepatocarcinogens in animals [14]. Regional differences in organochlorine pesticide use might explain regional differences in liver cancer mortality, but racial differences within regions might be difficult to explain on the basis of pesticide exposure. Factors that have not been addressed, such as the increasing industralization of the Southern regions and different employment patterns for whites and nonwhites might also
Liver Cancer
in the U.S.
235
influence HCC mortality trends. Thus, it is not possible to make definite conclusions from the data presented here. The apparent similarities in trends for cirrhosis and liver cancer mortality among nonwhite males suggests that cirrhosis may be a major risk factor for this group. For white males, among whom cirrhosis mortality is less common than among nonwhites and occurs at a later age, we must look to some other factor to explain the geographic differences observed. Such a factor might also play an etiologic role in the development of liver cancer in nonwhite males, but may be difficult to detect because of the apparently strong relationship to cirrhosis.
REFERENCES I.
2. 3. 4. 5. 6. 7. 8.
9. IO.
I I. 17. 13. 14. 15. 16. 17.
IX. 19. 70. 71. 22. 23. 24. 25. 26. 27.
28. 29.
Devesa SS. Silverman DT: Cancer incidence and mortality trends in the United States: 193551974. J Nat1 Cancer lnst60: 5455571, 1978 Saracci R. Repetto F: Time trends of primary liver cancer: indication of increased incidence in selected cancer registry populations. J Nat1 Cancer lnst 65: 241-247, 1980 Deichmann WB, MacDonald WE: Organochlorine pesticides and liver cancer deaths in the United States, 1930~-1972. Ecotoxic. Envir Safety I :89-l IO, 1977 Aoki K: Cancer of the liver, international mortality trends. World Hlth Stat Rep 3 I : 28-50, 1978 Blumberg BS. London WT: Hepatitis B virus and the prevention of primary hepatocellular carcinoma. N Engl J Med 304, 782-784. 1981 Szmuness W: Hepatocellular carcinoma and the hepatitis B virus: Evidence for a causal association. Prog Med Virol 24: 4&69, 1978 Zuckerman AJ: Hepatitis B virus and primary liver cancer. In Primary Liver Tumors. Remmer H, Bolt HM. Bannasch P, Popper H (Eds), Falk Symposium. Baltimore: University Park Press. pp. 4255430, 1978 Okuda K, Nakashima T: Hepatocellular carcinoma: A review of the recent studies and developments. In Progress in Liver Diseases. Popper H, Schaffner F (Eds). New York: Grune & Stratton, Vol. VI. pp. 6399650. I979 Shikata T: Primary liver carcinoma and liver cirrhosis. In Hepatocellular Carcinoma. Okuda K, Peters RL (Eds). New York: John Wiley. pp. 53-71. 1976 Hutt MSR: Epidemiology of human primary liver cancer. In International Agency for Research on Cancer. Liver Cancer IARC Sci. Pub. No. I, pp. 21-29, Lyon 1971 Omata M, Ashcaval M. Liew C: Hepatocellular carcinoma in the USA: Etiologic considerations. Gastroenterology 76: 279-287, 1979 Popper H: Hepatic cancers in man: Quantitative perspectives. Envir Res 19: 4822494. 1979 Stoloff L. Friedman L: Information bearing on the evaluation 01 the hazard to man from atlatoxin ingestion. PAG Bull 6: 21 37. 1976 Wogan GN: Aflatoxins and their relationship to hepatocellular carcinoma. In Hepatocellular Carcinoma Okuda K. Peters RL (Eds). New York: John Wilev, PP. 25542. 1976 Shank RC: Epidemiology of aflatoxin carcinogenksis.~ In Advances in Modern Toxicology Kraybill HF. Mehlman MA (Eds). New York: John Wiley. Vol. 3, pp. 291-318, 1977 Scheuer PJ : Pathopenesis and morbid anatomy of human primary liver cancer. In International Agency for Research on Cnncer. Liver Cancer. IARC Scientific Publication No. 1, pp. 13-20, Lyon, 1971 _ World Health Organization: Manual of the International Statistical Classification of Diseases, Iniuries. and Causes of Death,Vol. I. Based on the recommendations of the Seventh Revision Conference, 1955. Geneva, Switzerland, 1957 National Center for Health Statistics: International Classification of Diseases, adapted for use in the United States. DHEW, PHS Pub. No. 1693, 1968 National Center for Health Statistics: Vital Statistics of the United States: Mortality, 1958-1975. DHEW, PHS, Rockville. MD. Kleba AJ. Dolman AB: Comparability of mortality statistics for the seventh and eighth revision of the international classification of diseases. DHEW Pub. No. HRA/76-1340, Vital Statistics Series No. 2, 1975 U.S. Department of Commerce, Bureau of Census: 1970 Census of Population, Vol. 1, Characteristics of the population. Part 1. United States Summary. Section I. 1973 US Department of Commerce. Bureau of Census: Current Population Reports. Series P-25, Estimates of the population of the United States by age, race and sex. 1958-1975 US Center for Disease Control: Morbidity and Mortality Weekly Report, annual supplement. summary 1975. Vol. 24. No. 54. US DHEW No. (CDC) 76-8241, 1976 Hammond EC. Johnson H, Percy C: Reporting of cancer of the lung, liver and uterus. Cancer IO: 1802~1806, 1968 Percy C. Stanek E, Gloeckler L: Accuracy of cancer death certificates and its effect on cancer mortahty statistics. Am J Publ Hltb 71: 2422250. 1981 Devesa SS: Reliability of reported death rates and incidence rates. Prev Med 9: 5899600. 1980 Lehmann FG, Wegencr T: Etiology of human liver cancer: Controlled prospective study in liver cirrhosis. In Liver Carcinogenesis Lapis K, Johannessen JV (Eds). Washington: Hemisphere Publishing Corp, pp. I1 1~~129. 1979 Gurngliano CF. Lilienfeld AM, Mendeloff AI: Incidence rates of liver cirrhosis and related diseases in Baltimore and selected areas of the United States. J Chron Dis 32: 5433554. 1979 Noble EP: Third Special Report to the Congress on Alcohol and Health. Rockville, Maryland. US DHEW, National lnstitute on Alcohol Abuse and Alcoholism. 1978
DALE P.SANDLERettrl.
236
30. Breslow NE, Enstrom 31. 32. 33.
34.
35. 36. 37. 38. 39.
JE: Geographic correlations between cancer mortality rates and alcoho~tobacco consumption in the United States. J Nat1 Cancer Inst 53: 631-639, 1974 Keller A: Alcoholic beverages and carcinogenesis. NY State J Med 80: 1245-1250, 19X0 Mettlin C: Nutritional habits of blacks and whites. Prev Med 9: 601 -606, 1980 Klatsky AL. Friedman GD, Siegelaub AB, Gerard MJ: Alcohol consumption among white. black or oriental men and women: Kaiser-Permanente multiphnsic health examination data. Am J Epid 105: 31 l-323, 1977 Brechot C. Nalpus B, Courouce A, Duhamel G. Callard P. Carnot F, Tiollais P. Berthelot P: Evidence that hepatitis B virus has a role in liver-cell carcmoma in alcoholic liver disease. N Engl J Med 306: 1384-1387, 1982 Yarrish RL, Werner BG, Blumberg BS: Association of Hepatitis B virus infection with hepatocellular carcinoma in American patients. Int J Cancer 26: 71 I-715. 1980 Keen P. Martin P: Is aflatoxin carcinogenic in man’? The evidence in Swazil;lnd. Trap Geogr Med 23: 49-53. 1971 Alpert ME. Hutt MSR, Wogan GN: Association between aflntoxin contamination of food and hepatoma frequency in Uganda. Cancer 28: 253-260, 1971 Peers FG, Linsell CA: Dietary aflatoxins and liver cancer: A population based study in Kenya. Br J Cancer 27: 473-489, 1973 Bababunmi EA: Excretion of aflatoxin in the urine of normal individuals and patients with liver diseases in Ibudan (Nigeria). In International Study Group for the Detection and Prevention of Cancer. Prevention and Detection of cancer. Proc 3rd Int Symp Detection and Prevention of Cancer New York. Vol. 2. Part I, pp. 1729-1736, 1978