TOBACCO AND ALCOHOL CONSUMPTION IN RELATION TO THE DEVELOPMENT OF MULTIPLE PRIMARY CANCERS

ERNST WYNDER, MARGARET MUSHINSKI, L. MD, H. MA,

AND JOAN

C. SPIVAK, AB

The relationship between tobacco and alcohol consumption and the development of additional primary cancers of the upper alimentary tract is reviewed. The chance of developing a second primary is dependent principally on the intensity (i.e., quantity and duration) of the smoking and drinking habit prior to the onset of the 6rst neoplasm. However, results conflict regarding the effect exerted by the continuation of these habits after the first diagnosis. While tobacco smoking is considered the primary risk factor associated with cancers in this area, its interaction with alcohol creates a powerful carcinogenic effect. It is agreed that multiple primaries are selective on a site-specific basis and that risk varies with anatomic location of the first primary. Cancer 40:1872-1878,1977.

the development of cancer have reported that perhaps as much as 90% of all cancers are related to environmental factors. L,4420 Although estimates of this percentage vary considerably, it appears obvious that most cancers are induced by personal habits associated with lifestyle, as well as by other environmental factors. We can say with some confidence, therefore, that cancer is not a n inevitable consequence of aging and that many neoplastic diseases are amenable to preventive measures. Multiple primary cailcers have proved to be a n interesting area of epidemiologic investigation. Here, too, consensus is lacking, for estimates of the frequency with which second primary cancers occur vary from a high of 10.0" increased risk after the primary tumor occurrence to a low of 1.29.'' A partial reason for the variation in estimates is that studies of such diseases are subject to a number of biasing factors, including misclassification of the primary site of tumors in proximity to each other and metastatic diseases being considered as a second
Presented at the International Workshop on Multiple Primary Cancers, New York, New York, October 7-8, 1976. From The American Health Foundation, New York, New York. Supported in part by contract from National Cancer Institute, SHP-74-104, and in part by National Cancer Institute Grant Nos. CA-17613 and CA-12376. Address for reprints: E. L. Wynder, MD, American Health Foundation, 1370 Avenue of the Americas, New York, NY 10019. Accepted for publication May 17, 1977.

ARIOUS EPIDEMIOLOGIC INVESTIGATIONS ON

primary neoplasm. It is agreed, however, that multiple primaries are highly selective on a sitespecific basis and that risk varies with anatomic location of the first primary. Goodner and Watson reported that half of the 126 second primaries in esophageal cancer patients were of the head and neck area and half were of the intraoral cavity.' As early as 1933, Lund reported that more than twice the expected number of patients with a primary oral cavity cancer died of a second primary.' These data indicate that a predisposing factor active in the causation of esophageal cancer may also play a role in the incidence of intraoral cancer. Using a simple life-table method to predict the occurrence of a second primary cancer, Epstein concluded that the presence of malignant neoplasms in the respiratory and upper digestive tracts implies an enhanced risk for the occurrence of another malignancy in that same ana. tomic region.' In addition, Cahan found that persons with primary lung cancers frequently developed a second cancer in the oral cavity or larynx.' Multiple primary cancers have been studied both with regard to site of occurrence and to the influence exerted by prior and continued smoking and drinking habits. For example, it has been found that patients with cancer of the mouth, pharynx and larynx often develop a second cancer in the same region. Existing evidence shows that excessive tobacco and alcohol usage are the major etiologic factors for the development of cancers in these areas. In addition, data

1872

No. 4

MULTIPLE PRIMARY CANCER TOBACCO-ALCOHOL Wynder et al. AND

9m-

1873

"I
I)*
I I

Ym

8
FIG.1. Tumor response to different doses of cigarette smoke condensate in acetone.

......... .. ..... .... ..

i

3 B

i::
O i
; '5
6

...........
,
14

" I . . . "

......
0-

10

'

/ , . , ,#

,
17

, % 1I

11 12 MONTHS

13

I5

16

have also shown that those at greatest risk of a second primary cancer are those who smoke the most. Before discussing the effects that cessation or continuation of smoking and/or drinking exert on the development of a second primary cancer, it is relevant to discuss the independent influence of these factors on primary cancers. It is no surprise that experimental evidence has shown that the greater the potency exerted by an agent and the longer the duration of exposure to that agent, the greater will be the chance of developing at least one anomaly in the host. That is, a .005% solution that produces a tumor late in an animal's life is less likely to produce a second tumor in that animal; but the chances of developing a second primary are greatly enhanced if you expose an animal to a 0.5% solution earlier in life and expose it over a longer period of time. Thus, dose and duration are seen to increase the frequency of occurrence of both benign and malignant tumors and effect a progressive increase in the number of tumorbearing animals being treated (Fig. 1). An analogous situation occurs in man with respect to cancers of the upper digestive tract and oral cavity. It has been shown that multiple primaries of these systems are related to man's exposure to such substances as alcohol and tobacco, over a relatively long life span. Epidemiologic studies have proved that these two factors are major etiologic agents in the development of cancer of the upper alimentary tract and that duration and quantity of tobacco exposure significantly increase the risk of disease. The association between tobacco exposure and risk of various cancers is well known and is demonstrated by an almost linear dose-response relationship between amount smoked and the

development of a variety of tobacco-related cancers (Fig. 2). Although the increase in risk is not as dramatic for cancer of the esophagus, oral cavity, or urinary bladder as it is for lung and larynx, it is clear that in the absence of smoking,

81

ESOPHAGEAL CANCER

BLADDER CANCER

No. OF CIGARETTES PER MI

Flo. 2. Relative risk for six tobacco-related cancers for males, by amount smoked.

1874

CANCER October Supplement 1977

LUNG CANCER ( I I
N 3T1 CMES 812 CONTROLS

vol. 40

1-10

11-20

21-10

31-40

41 t

LARYNX CANCER

1-10
WF 1976

11-20

21-30

31-40

41+

Na CIGARETTES WORE0 PER M Y

FIG.3. Relative risk of cancer of lung and larynx among male smokers.

the occurrence of these cancers would be quite rare." Although studies have shown a redhced risk for many of these cancers in long-term smokers of filter cigarettes, as compared with smokers of nonfilter cigarettes, the risk remains considerably greater than that of nonsmokers (Fig. 3). Cigar atld pipe smoking has also been implicated in the etiology of oral cavity and esophagus cancer, but to a lesser extent than cigarette smoking (Fig. 4). That cigar and pipe smoking does not appreciably increase the risk of cancer of the lung or larynx is an indication that the key variable here is the degree of inhalation; the cigarette smoker consciously inhales more often and more deeply than the cigar or pipe smoker. That the combined effect of tobacco and alcohol is of greater significance than the effect of alcohol alone is borne out by a recent study, which found no ihcreased risk for larynx cancer among nonsmoking heavy drinkers, although heavy drinking in combination with tobacco usage was found to increase risk significantly, especially for cancer of the supraglottis (Fig. 5). Thus, while tobacco is the prime risk factor in the development of cancer of these areas, its interaction with alcohol creates a very powerful carcinogenic effect. I' " The risk of developing cancers of the lung, larynx, pharynx, oral cavity and esophagus is known to decrease with the cessation of tobacco

4
Y

PIPE, N.38

v, a

6.

> a I
i=
a

IA

4 .

5J
24:509
SMOKERS

74

FIG.4. Relative risk of cancer of the oral cavity for men.

30 .4

.0.

I-5

6-10

CIGARS OR PIPE BOWLS PER DAY

No. 4

MULTIPLE PRIMARY CANCER TOBACCO-ALCOHOLWynder et al. AND

1875

1-6

FIG.5. Relative risk of laryngeal d canFer for male smokers by amount 2 0 smoked ahd alcohol consumed. a 1 * Relative to the risk of 1.0 for non- > smokers who never or only occasionally drank alcohol. % i

*I
-J

n rr + ' c

ALCOHOL UNITS PER DAY

n r

;a
1

1-15
usage. l6 It has been shown that after 15 years of nonsmoking, the risk of developing lung cancer is almost that of someone who never smoked (Fig. 6). Likewise, there is a similar decrease in risk with cessation of smoking for cancer of the larynx. I' This fact is particularly relevant to a discussion of second primary cancers because it must be recognized that once a cancer cell is initiated, it frequently may take between 1 and 10 years before the clinical appearance of the cancer. Since the relative risk of developing cancer remains fairly stable for at least 3 years after
5081

16-34 TOBACCO UNITS SMOKED DAILY

smoking cessation, the effect produced by exsmoking on the development of a second primary cancer is almost impossible to measure prior to this interval. In addition, since the interval between the occurrence of the first and second primary is sometimes as short as 1 year, the synergistic effect of smoking and alcohol is still present. The second major risk variable in the development of cancer of the larynx, oral cavity and esophagus is heavy consumption of alcohol lS (Fig. 7). It is important to note that there is
LARYNX CANCER, MALES
CONTROLS CASES

LUNG CANCER I.MALES

254
107

CONTROLS CASES

syO*Ern YEARS ff IMOMINO QIIlTlON

SMOKERS

2%

FIG.6. Relative risk of cancer of the lung and larynx, for males and femaies, by years of smoking cessation.

zz

SYQERS ~ffSyoII1IGQssITlow

I 4

1 0
5 -

Y E I I I O C )(Io*waATlOW

-m

*K)llEm
YLAMQIY0010C-OW

smms

1876

CANCER October Supplement 1977

ORAL CAVITY CANCER

Vol. 40

Drinks 81 per diiy Alltypbs

1-6

7+ Whiskey

1-6

7+ Beer, Wine

ESOPHAGEAL CANCER'

Drinks per day

Whiskey

1-6

7+

Cl

Beer

1-6

7+

Wine

EXTRINSIC LARYNGEAL CANCER"

Drinks per day

1-6
Whiskey

7+

Beer Wine

Fro. 7. Relative risks of developing cancers of the oral cavity, esophagus, and larynx by alcohol consumption. * Relative risk standardized for tobacco consumption. ** Relative risk compared with those smoking 16-34 cigarettes/day.

neither epidemiologic nor experimental evidence that alcohol, per se, is carcinogenic. However, in combination with cigarette smoking, excessive alcohol intake significantly increases the risk for these cancers. Studies have shown that the relative risk of disease increases proportionally with type of alcohol when standardized for tobacco usage. That is, excessive consumption of hard liquor (i.e., 7+ units per day) by a heavy cigarette smoker creates a situation of highest risk for the individual, a risk which decreases at lower levels of consumption and with type of beverage consumed. Therefore, there appears to be a synergistic effect between tobacco and alcohol, with alcohol acting as a promoter. Nutritional deficiencies, commonly associated with alcoholism, may be the integral factor in the mechanism for this relationship. Studies have proved that the occurrence of the second primary is dependent on the patient's

smoking and drinking history prior to the first primary. It was not until Moore's work in 1965, however, that efforts were directed towards analyzing smoking habits of patients with cancer of the mouth after onset of illness. lo He found that the risk of developing a second primary cancer in patients who continued smoking after diagnosis of the first primary in a tobacco-critical region was greatly enhanced. The study showed that 20 out of 65 patients who continued smoking developed a second primary, as opposed to only 2 out of 37 patients who had stopped smoking. Moore did not study the incidence in relation to the various forms of tobacco used (i.e., cigarettes, cigars or pipes) because he felt their effects did not differ significantly. The implication of this study is that patients treated for cancer in the tobacco-critical regions could generally prevent the occurrence of a second primary in those areas by cessation of tobacco use. In direct contrast to Moore's study is one by Castigliano, who concluded that smoking cessation after the development of the first primary does not decrease the chances of developing a second primary. He found that the incidence of second primaries was the same regardless of the patient's past or present smoking habits, or lack of them.' In a retrospective study, Wynder et al. found that the more a n individual smoked before the development of a first primary, the greater was the chance of developing a second primary. l5 This result obtained even for those who gave up smoking after diagnosis of the first primary. Moreover, the interval between development of first and second primary lesions has been found to be related to amount of tobacco consumed previous to diagnosis of the index cancer. That is, longer intervals ( 2 1 0 years) were found for those who smoked the least than for those who smoked the most (S4 years). However, these differences were not significant. *' It was also found that continued smoking exerted a greater influence on the development of a second primary than did continued drinking. Similar results were found by Schottenfeld et al. in a prospective study of 733 patients with primary cancer of the oral cavity, pharynx and larynx. l3 Previous quantity smoked and, more significantly, the combined effect of high levels of cigarette and alcohol consumption prior to diagnosis of the index cancer led to an increased relative risk of developing a second primary 3.6 times that of patients with lower levels of alcohol and tobacco exposure. Berg et al. found the rate of occurrence of

No. 4

MULTIPLE PRIMARY CANCERN D TOBACCO-ALCOHOL A

Wynder et al.

1877

second primaries to be 150-250% greater than expected in 9415 patients, with the exception of index patients with cancer of the nasal cavity, paranasal sinuses and nasopharynx. Although patients with oral cancer developed almost one new cancer of the esophagus, larynx or lung for every new primary oral cancer (100/116), it was found that different intraoral sites for the first cancer were associated with significantly different risks for specific types of later cancers. Two factors were ascribed to the excess risk for multiple primaries in this study. The first was field carcinogenesis-the development of dysplastic changes somewhat asynchronously over a general epithelial region. The second factor was the exposure of the patient to tobacco. Although detailed smoking histories and follow-ups were outside the realm of this study, medical records confirmed long smoking histories for almost all patients. Alcoholism was also frequently mentioned in the history, although somewhat surprisingly, more often for women than men. In a later study by Moore, it was found that of patients with index cancers of the oral cavity, pharynx or larynx, 40% who continued to smoke developed a second primary, while only 6% who stopped smoking after onset of the first lesion did so. The death rate for contined smokers was higher and occurred earlier than in those who stopped smoking. Another issue that obviously relates to the risk of developing a second primary is the survival rate for the first cancer. Not only do the rates of survival differ between various anatomic sites, but they are also influenced by the extent of the lesion and the treatment regimen prescribed for the patient (Fig. 8). All of these factors must be considered when determining the risk for another cancer. For example, because the survival rate for a primary cancer of the esophagus is quite low, ones chance of developing a second primary in the tongue is minimized. This is similar for cancer of the pharynx, which has an average survival time of 2.5 years. On the other hand, if the first primary is in the larynx, which has a relatively high rate of survival, the chance of developing a second primary elsewhere later in life will be increased. Although not the tvpic of discussion in this presentation, it should be noted that of additional etiologic significance in the development of a second primary cancer is the possibility of radiation therapy for one cancer acting as a predisposing factor for the occurrence of another cancer. For instance, the 1969 study by Wynder et al. also found that the risk of a second primary

------. .._..._ Intrinsic

hryna

BUCUl

mucow
Gum

i
2 v*
1

ka % Pm

mth

Tonsil Edrinrk

5
VW5

l k

1 0

FIG.8. Cancer mortality, Memorial Hospital, 1949-1957.

after the index cancer was related to treatment of the first primary with radiation. l6It appeared that the major cause of a second primary was the amount of radiation used to treat the initial lesion. In summary, we have shown that certain environmental and lifestyle factor are important for the induction of cancers of specific sites. Tobacco usage, in particular, has been identified as causative for cancers of the lung, larynx, oral cavity, esophagus, and urinary bladder. Several investigators have observed that patients with cancers of these sites developed a second primary cancer within the same anatomic region. Moreover, it appeared that continuation of smoking habits after diagnosis of the index cancer increased ones risk for development of a second primary lesion. Heavy alcohol consumption also predisposes certain sites to tumorigenesis, but primarily in conjunction with tobacco usage. We have also shown that tobacco-related cancer sites and those related to alcohol consumption will be less prone to the development of second primaries if patients smoked and drank less. Greater activity in the area of antismoking campaigns, smoking cessation programs, and the development of the less harmful cigarette

1878

CANCER October Supplement 1977

vol. 40

will, hopefully, result in the reduction of multiple primary tumors, and tobacco-related cancers as well. Epidemiologic and experimental studies of multiple primary cancers are important and should be continued. Greater understanding of the mechanism involved, i.e., tissues and organs

with increased epithelial susceptibility and the specific environmental conditions which are of etiologic significance in the development of multiple carcinomas, is in order. Such investigations might lead to the day when detection of high risk persons can be maximized and the occurrence of second primary cancer minimized.

REFERENCES
1. Armstrong, B., and Doll, R.: Environmental factors and cancer incidence and mortality in different countries with special reference to dietary practices. Int. 3. Cancer, 15:617-631, 1975. 2. Berg, J. W., and Schottenfeld, D.: Incidence of multiple primary cancers. 111. Cancers of the respiratory and 12. Schoenberg, B. S.: Multiple primary neoplasms. In Persons At High Risk of Cancer: An Approach to Cancer Etiology and Control, J. F. Fraumeni, Jr., Ed. New York, Academic Press, 1975; pp. 103-120. 13. Schottenfeld, D., Gantt, R. C., and Wynder, E. L.: The role of alcohol and tobacco in multiple primary cancers of the upper digestive system, larynx and lung-A prospece. tive study. R v Mcd. 3:277-293, 1974. 14. Schottenfdd, D., and Berg, J. W.: Epidemiology of multiple primary cancers. In Cancer Epidemiology and Prevention, D. Schottenfeld, Ed. Springfield, Ill., Charles C Thomas, 1975; pp. 416-436. 15. Wynder, E. L., Dodo, H., Bloch, D. A., el al. Epidemiologic investigation of multiple primary cancer of the upper alimentary and respiratory tracts. I. A retrospective study. Cancer 24~730-739, 1969. 16. Wynder, E. L. and Stellman, S. D., The comparative epidemiology of tobacco-related cancers, submitted to Cancer Research. 17. Wynder, E. L., Covey, L. S , Mabuchi, K., and . Mushinski, M. H.: Environmental factors in cancer of the larynx-A second look, Cancn 38:1591-1601, 1976. 18. Wynder, E. L., and Stellman, S. D.: The epidemiology of the less hazardous cigarette-A retrospective study of 1399 cases of lung and larynx cancer. Submitted to JAMA 19. Wynder, E. L., and Mabuchi, K.: Etiological and e. preventive aspects of human cancer. R v Mcd. 1:300-334, 1974. 20. Wynder, E. L., and Gori, G. B.: The contribution of

upper digestive system as multiple primary cancers. J . Natl. Cancer Inst. 443264274, 1970. 3. Cahan, W. G.: Lung cancer associated with cancer primary in other sites. Am. J. Surg. 89:494-514, 1955. 4. Castigliano, S. G.: Intluence of continued smoking on the incidence of second primary cancers involving mouth, pharynx and larynx. J . Am. Dmf. Assoc. 77:580-585, 1968. 5 . Doll, R.:Strategy for detection of cancer hazards to man. Nature 265589-596, 1977. 6. Epstein, S. S., Payne, P. M., and Shaw, H. J.: Multiple primary malignant neoplasms in the air and upper food passages-A statistical review. Canctr 13:137-145, 1960. 7. Goodner, J. T., and Watson, W. L.: Cancer of the esophagus-Its association with other primary cancers. Cancer 9:1248-1252, 1956. 8. Higginson,J. In Environmental Pollution and Carcinogenic Risk: IARC Scientific Publication No. 13. Paris, INSERM, 1976. 9. Lund, C. C. : Second primary cancer in cases of cancer of buccal mucosa-Mathematical study of susceptibility to cancer. N.Engl. 3. Med. 209:1144-1152, 1933. 10. Moore, C.: Smoking and cancer of the mouth, pharynx and larynx. JAMA 191:283-286, 1965. 11. Moore, C.: Cigarette smoking and cancer of the mouth, pharynx and larynx-A continuing study. JAMA 281:553-558, 1971.

the environment to cancer incidence-An epidemiologic mercise. 3.Natl. C n e Insf., 58:825-831, 1977. acr