Cigarette smoking and leukemia

J Cbmn Dis Vol. 39, No. 6, pp. 417421, 1986 Printed in Great Britain

CIGARETTE

0021-9681/86 53.00 + 0.00 Pergamon Journals Ltd

SMOKING

HARLAND

AUSTIN

AND LEUKEMIA*

and

PHILIP

COLE

Department of Epidemiology and the Comprehensive Cancer Center, University of Alabama in Birmingham, Birmingham, AL 35294, U.S.A. (Received in revised form 10 December

1985)

INTRODUCTION THE REPORT of the Surgeon General of the United States cites cigarette smoking as a major cause of cancers of the lung, larynx, oral cavity, and esophagus, and as a contributory factor in the etiology of cancers of the bladder, kidney, and pancreas [ 11.It is also indicated in this report that cigarette smoking may be related to stomach and cervical cancer. Leukemia is not considered a tobacco-related cancer, although the report mentions that several epidemiologic studies have reported higher leukemia risks among smokers compared with non-smokers [2]. This evidence was judged inconclusive because no dose-response relationship between leukemia mortality and the amount smoked was found in these studies. This paper reviews briefly the epidemiologic evidence pertinent to an evaluation of the relationship between cigarette smoking and leukemia and suggests strategies for the further evaluation of this relationship.

EPIDEMIOLOGIC

STUDIES

In a review of the epidemiology of leukemia, Alderson concluded that cigarette smoking was not an etiologic factor [3]. This conclusion was based on the lack of an association between smoking and deaths from cancers of the blood and reticuloendothelial system in the prospective follow-up study of British male doctors [4]. However, there are several other epidemiologic studies which report data on cigarette smoking and leukemia. An appraisal of these studies suggests that Alderson’s conclusion was premature. Prospective follow-up studies

The results of three large prospective follow-up studies of cigarette smoking and leukemia mortality appear in Table 1. Each of these studies is generally considered to have been well designed and well executed. In the British doctors study, there is a deficit of leukemia deaths among current or ex-smokers compared with non-smokers. Furthermore, among smokers there is an inverse relationship between leukemia mortality and the amount of tobacco smoked daily. On the other hand, in the two follow-up studies from the United States, there is an overall excess of leukemia mortality of about 50% among cigarette smokers compared with non-smokers. In the United States Veterans study, there is some evidence of a dose-response relationship among current smokers, but the trend is errratic and is not statistically significant (one-tailed p value = 0.19). There are two other prospective follow-up studies relevant to an evaluation of the relationship between cigarette smoking and leukemia mortality. In a study of male labor *Supported by grants from the National Institute.

Cancer Institute (S-30-CA13148) and the American Petroleum

417

Ex-smokers: 119 Curreot smokers: 207

Expected number of leukemia deaths’

1.6

1.5

1.5(1.1,1.9)

=

Mortality rate ratio 1.0 1.9(1.4,2.6)

1524 22

Mortality rate ratio 1.0 1.2 1.6 1.5 0.002)

Not Reported

Current cigarettes/day Non-smoker ~< IO IO-20 21+ Ys-* = 2.9 (P,

1.5(1.1,2.1)

.~

current cigarettes/day Non-smoker 20+

1-14 27

Dose-response evaluation

Grams/day3 Mortality Rate

1.8(1.4,2.4)

0.7(0.5, 1.1)2

Mortality rate ratio

‘Deaths from cancers of the bone marrow and reticuloendothelial system. ‘Approximate 95% confidence interval. )Number of grams per day of any kind of tobacco among current smokers. ‘Young includes ages 4W; old includes ages 6579. ~Expccted number of leukemia deaths based upon the leukemia mortality experience of non-smokers.

Current Smokers: 333

Ex-smokers: I75

16

Current smokers: 23

U.S. Veterans (16 years of follow-up) [8]

16

U.S. Veterans (8 years of follow-up) [7j

29

Ex-smokers: 24

300

8

Young:’ 13 Old:’ 57

33

Non-smokers

rates (IO:)-’

311

152’

24

Smokers

Mortality

American cancer Society [S, 61

British Doctors 141

Study

Leukemia deaths (men only) 25+ 19

Cigarette Smoking and Leukemia TABLE

419

2. THE RELATIVE ~NCDENCE RATES OF ACUTE AND CmoNrC

MYELOC’YTIC LEUKEMIAS ACCORDING

TO PACK-YEARS OF CIGARETTE

SMOKING AND GENDER (THIRD NATIONAL CANCER SURVEY INTERVIEW STUDY)

Chronic Pack-Years <20 20-39 40+ Total Cases

Myelocytic

Men

Acute Myelocytic

Women

Men

Women

-_(O) 3.2(S)

0.8 (I) 3.3 (2) 2.6(l)

1.6 (5) I .4 (3) I. 1(4)

1.6 (2) 8.8 (4) 2.6(l)

IO

I2

27

23

1.8’.*(2)’

‘The reference value is I .O for non-smokers. 2Adjusted for differences between cases and controls ‘Number of cases in that category.

in age and race.

union members in California in which 30 leukemia deaths occurred, the relative rate (RR) of leukemia mortality for cigarette smokers compared with non-smokers was 1.3 [9]. In a Japanese follow-up study which included 33 male leukemia deaths, the RR of leukemia mortality for daily cigarette smokers compared with non-smokers or occasional smokers was 0.8 [lo]. However, these two latter studies are small and are not inconsistent with a small or moderate increase in leukemia risk attributable to cigarette smoking. Case-control siudies Two case-control studies are also relevant to an evaluation of the relationship between cigarette smoking and leukemia. In the interview study from the Third National Cancer Survey, leukemia cases were reported according to chronicity and cell-type [1 11. Controls in this study were persons with cancers of various sites which are not generally believed to be related to cigarette smoking. The RR’s of acute and chronic myelocytic leukemia according to pack-years of cigarette smoking and gender are displayed in Table 2 (the referent category is non-smokers). The investigators reached the reasonable conclusion that there was a “suggestion” of a positive relationship between cigarette smoking and both types of myelogenous leukemia. There was no such positive relationship between cigarette smoking and lymphocytic leukemia. Paffenbarger et al. followed about 50,000 men who had attended Harvard University or the University of Pennsylvania between 1916 and 1950 [12]. After 1.7 million person-years of observation, 27 and 41 men had died from lymphatic and myeloid leukemia, respectively. For each of these decedents, four controls were chosen from among classmates born in the same year and who were known to have survived the decedent case. The RR of lymphocytic leukemia for cigarette smokers compared with non-smokers was 1.3 (95% confidence limits: 0.5, 3.2), while for smokers of more than 10 cigarettes per day, this RR was 2.7 (0.7, 11.2). For myelocytic leukemia, the RR for all cigarette smokers compared with non-smokers was 2.4 (1.1, 5.3), while for heavy smokers, the RR was 3.6 (1.1, 11.4). Interpretation of the epidemiologic studies None of the studies described above was designed specifically to evaluate the relationship between cigarette smoking and leukemia. As a result, they have major limitations with respect to an evaluation of this hypothesis. For example, the follow-up studies did not distinguish the acute from the chronic forms of leukemia, nor did they distinguish among the various cell-types of leukemia. There is some evidence that the different forms of leukemia have different etiologies, and it is possible that cigarette smoking is related to one, or several, but not to all forms of the disease. To the extent that this is true, these studies would present an underestimate of a true positive association of smoking with some particular form or forms of leukemia. The case-control studies do not suffer this limitation, but they are small and obtained only sparse information on cigarette smoking. Both the American Cancer Society and the U.S. Veterans studies suggest a small excess leukemia risk attributable to cigarette smoking. The inference that cigarette smoking causes leukemia is supported by the results of the two case-control studies. More

420

HARLAND AUSTIN and PHILIP COLE

specifically, these studies suggest that smoking is related to myelogenous leukemia. On the other hand, the British Doctors study does not support the inference that cigarette smoking causes leukemia. Although this study has only about one-half the number of leukemia deaths as does each of the American follow-up studies, it is difficult to attribute the absence of a positive smoking effect among British Doctors to chance since there was an inverse trend between leukemia mortality and amount of tobacco smoked which was almost statistically significant (p = 0.06). The discrepancy between the British and the American studies may provide an opportunity to identify the leukemogen in cigarette smoke, if such an agent exists. It is possible that American cigarettes contain a leukemogen which is absent from the cigarettes smoked in Britain.

BIOLOGIC

PLAUSIBILITY

Many carcinogenic agents have been identified both in the gas phase and in the particulate matter of cigarette smoke [2]. For example, cigarettes contain urethane and nitrosamines which are leukemogenic in experimental animals [13-181. Of particular interest to the cigarette smoking-leukemia hypothesis is the benzene and radioactive compounds in cigarettes, since both of these agents cause leukemia in man. The smoke of one cigarette contains between 10 and 100 pg of benzene [2]. If an average of 30 pg of benzene per cigarette is assumed, a person smoking one pack of cigarettes daily would be exposed to about 600 pg of benzene from this source. Although this amount of benzene represents about a 67% increase over the “baseline” level of benzene exposure from water, air, and food [19], it is a small amount compared with what would be sustained by a worker exposed to the current occupational standard in ambient air of 32 pg of benzene/liter. Such workers would be exposed to about 300 times the amount of benzene sustained by a person smoking a pack of cigarettes each day. Yet, it is unlikely that these workers experience more than a five-fold increase in leukemia risk above that of the general population. Thus, it is not reasonable to attribute a two-fold increase in leukemia risk among smokers solely to the benzene content of cigarettes. Cigarette smoke contains radioactive elements that could cause an increased risk of leukemia among smokers. Of particular concern are the a-emitting “‘PO (polonium) and its long-lived p-emitting precursor z’“Pb (lead). Higher concentrations of these elements have been found in the liver, kidney, spleen, pancreas, and lungs of smokers compared with non-smokers [20,21]. Holtzman reported a mean level of 0.14 pCi of *‘OPbper gram of rib bone ash among six non-smokers, while among 13 smokers the mean was 0.28 pCi per gram of bone ash [22]. He estimated that this doubling of the “‘Pb could increase the total skeletal dose by as much as 30%. It is interesting that this author speculated that this source of excess radiation might place smokers at an increased risk of leukemia. The presence in cigarettes of chemical and radioactive agents known to cause leukemia in animals or man provides some biologic credibility to the hypothesis that cigarette smoking is a cause of leukemia. These substances could act in conjunction with one another to increase leukemia risk among smokers. Since cigarette smoking is a known cause of cancer of organs not in direct contact with the smoke (bladder, kidney, and pancreas), it is apparent that some carcinogen(s) in cigarettes reach these sites. The presence of known chemical leukemogens in cigarette smoke suggests that cigarette smoking may also cause leukemia. SUMMARY

AND

RECOMMENDATIONS

The epidemiologic evidence cited above provides some support for the inference that cigarette smoking is a risk factor for leukemia. Yet, it is not possible to conclude that this evidence reflects a “valid and causal” relationship. This is so because the relationship is weak and not all of the studies are positive. In particular, the British Doctors study is inconsistent with the hypothesis that cigarette smoking is a cause of human leukemia. On

Cigarette Smoking and Leukemia

421

the other hand, other epidemiologic studies are suggestive of a positive association and we are not able to suggest any credible sources of bias which would explain the reasonably consistent positive effect seen in these studies. Although the epidemiologic evidence pertaining to the cigarette smoking-leukemia hypothesis is currently equivocal, it is our opinion that there is enough of a suggestion of a positive association to warrant further investigation. The relationship between cigarette smoking and leukemia might be clarified by more detailed analysis of existing data sources. As mentioned above, the reports of the prospective follow-up studies did not distinguish the various forms of leukemia. Experimental and epidemiologic evidence suggest that smoking is more likely to be related to AML rather than to other forms of leukemia. If these published reports could be re-analyzed to provide estimates of leukemia mortality rates according to form of leukemia and smoking status, much might be learned about the smoking-leukemia hypothesis. In particular, if it were found that AML is more strongly associated with smoking than are other forms of leukemia, the more specific inference that cigarette smoking is a cause of AML would be strengthened. It may be possible to evaluate the relationship between smoking and specific forms of leukemia in extant caseecontrol studies. In future case-control studies of the leukemias, a moderately detailed history of cigarette smoking should be obtained. Until such data are available, it will remain speculative that cigarette smoking is a cause of leukemia. REFERENCES I.

U.S. Department

of Health Services: The health consequences of smoking. Cancer. A Report of the D.C.: U.S. Govt. Print. Office, 1982. [DHHS (PHS) 82-501791 U.S. Department of Health, Education, and Welfare: Smoking and Health. A Report to the Surgeon General. Washington, D.C.: U.S. Govt. Print. Office, 1979. [DHEW (PHS) 79-500661 Alderson M: The epidemiology of leukemia. Adv Cancer Res 31: I-76, 1980 Doll R, Peto R: Mortality in relation to smoking: 20 years observations on male British doctors. Br Surgeon General. Washington,

2. 3. 4.

Med J 2: 1525-1536, 1976

5. 6.

7. 8. 9. IO.

11. 12. 13. 14. 15.

Hammond EC: Smoking in relation to death rates of one million men and women. Nat1 Cancer Inst Monogr 19: 127-204, 1966 Hammond EC: A multiple factor analysis of the association between cigarette smoking and mortality. (Unpublished.) A paper read at a meeting of the American Association for the Advancement of Science on December 28, 1971 Kahn HA: The Dorn study of smoking and mortality among U.S. Veterans: Report on 8.5 years of observation. Nat1 Cancer Inst Monog 19: i-125, 1966 Rogot E: Smoking and mortality among U.S. Veterans. J Chron Dis 27: 189-203, 1974 Weir JM, Durn JE: Smoking and mortality: A prospective study. Cancer 25: 105-112, 1970 Hirayama T: Prospective studies on cancer epidemiology based on census population in Japan. In: Cancer Epidemiology, Environmental Factors. Bucalossi P, Veronesi U, Cascinelli N (Eds). Vol 3. Proc Xl International Cancer Congress, Florence, October 2&26, 1974. Amsterdam: Excerpta Medica, 1975. pp. 2636. Williams RR, Horm JW: Association of cancer sites with tobacco and alcohol consumption and socioeconomic status of patients: Interview study from the Third National Cancer Survey. J Nat1 Cancer Imt 58: 525-547, 1977 Paffenbarger RS, Wing AL, Hyde RT: Characteristics in youth predictive of adult-onset malignant lymphomas, melanomas, and leukemias: Brief Communications. J Nat1 Cancer Inst 60: 89-92, 1978 Kawamoto S, Kirschbaum IA, Taylor G: Urethan and leukemogenesis in mice. Cancer Res 18: 725-729, 1958 Fiore-Donati L, Chieco-Bianchi L, DeBenedictis G et al: Leukaemogenesis by urethan in new-born Swiss mice. Nature 190: 278-279, 1961 Vesselinovitch SD, Mihailovich N: Significance of newborn age and dose of urethan in leukemogenesis. Cancer Res 26: 1633-1636,

1966

16. Shisa H, Matsudaira Y, Hiai H et al: Origin of leukemia cells in mouse leukemia induced by N-butylnitrosourea. Gann 66: 3742, 1975 17. Odashima S: Leukemogenesis of N-nitrosobutylurea in the rat. I. Effect of various concentrations in the drinking water to female Donryu rats. Gann 61: 245-253, 1970 18. Hadjiolov D: Thymic lymphoma and myeloid leukemia in the rat induced with ethylnitrosurea. Z Krabaforaeh 77: 98-100, 1972 19. Newbeme P: Known or suspected chemical carcinogens in cigarette smoke. A report prepared for the American Petroleum Institute. (Unpublished.) October 24, 1983 20. Blanchard RL: Concentrations of *i”Pb and 2’oPo in human soft tissues. Health Phys 13: 625432, 1967 21. Hill CR: Polonium-210 in man. Nature 208: 423428, 1965 22. Ho&man RB, Ilcewicz FH: Lead-210 and Polonium-210 in tissues of cigarette smokers. Science 153: 1259-1260, 1966