Document bOaGkMr5ay37pXgB9GKMVoEvo

CHAPTER THIRTY-TWO Lung and Pleura Joseph F. Fraumeni, Jr. William J. Blot INTRODUCTION below 25 per 100,000; vet even in these areas lung cancer still ranks as one of the three most In "'e Lnited Slates primary lung cancer common cancers among males. Based on avail represents apout 1 5 per cent of ail cancer cases able international statistics and trends, it seems 02 per cent in males and 8 per cent in females). lively that lung cancer now surpasses stomach Because ot high fatality rates, the disease ac cancer as the world's most common malignant counts tor 25 per cent of all cancer deaths (34 neoplasm among males. The incidence is gen per cent in males and 1 5 per cent in females). It erally 4 to 6 times higher in males than females, is clear that the predominant risk factor is whose reported rates are usually under 25 per cigarette smoking, with industrial exposures 100,000. However, the male to female ratio is also plaung an important role, so that lung much lower in some groups, including Chinese cancer i, a largely preventable disease. This populations, some Latin American countries, chapter reviews the epidemiologic and etiologic and the Maoris of New Zealand. In large part aspects ot lung cancer, including the com the worldwide variation and male predomi paratively rare mesotheliomas, which arise from the pleural (or peritoneal) lining usually as a consequence ot asbestos exposure. nance can be attributed to differences in tobac co consumption. Although less striking than the international patterns, variation in lung cancer has also been LUNG CANCER documented within countries. In the United States, lung cancer mortality during the period 1950-69 was elevated in urban areas of the Demographic Factors north, but the highest rates were clustered in the south, particularly in seaboard communities Geographic Variation along the Atlantic and Gulf coasts (Blot and Fraumeni, 19761. Since 1970, the excess mor Lung cancer is the leading form of cancer in tality has become more pronounced in southern many countries, particularly in western Europe areas, with only slight clustering in the urban and North America, with annual age-adjusted northeast (Fig. 1). Distinctive patterns are also incidence rates among males exceeding 100 seen in China, where mapping of mortality per 100,000 population in Great Britain, United statistics for 1973-77 has revealed high lung States bracks, and Finland (Waterhouse et al, cancer rates around the industrialized centers of 1976) (Table 1). The incidence is comparatively Shanghai, Tientsien, and Peking (Li and Shiang, low in China, India, and several Latin American 1980). Surveys in japan, Norway, Italy, and and African countries, where rates are near or several other nations also have indicated siz- 564 lung and PLEURA -- 565 TABLE 1. Lung Cancer Incidence Among Males in Selected Countries. 1970' Annual Age-Adjusted Rate <European Standardf Per 100.000 Population <25 25-49 50-74 75-99 too-*- Puerto Rico Iceland India Singapore (Malay. Indian) China Nigeria Israel U S (Hispanicl U S (Utah) Spam Norway Sweden Japan Brazil Colombia Poland Cuba Yugoslavia Hungary Canada U S (white) New Zealand (Maori) Germany Switzerland Singapore (Chinese) United Kingdom Finland U S (black) Prom Cancer incidence in Five Continents. Vol 3 (Waterhouse et al. 19761 The data were often derived from selected areas within the countries and do not necessarily represent national rates The time periods covered varied among the reporting registries, but usually centered about 1970 able variation1; within countries, with generally higher rates in urban areas and coastal commu nities. Time Trends and Age Curves Lung cancer is rapidlv increasing in most areas ot the world, following by about 20 sears a parallel trend in cigarette smoking. In Great Britain, lung cancer increased 50-fold in men, from an age-standardized death rate of about 2'10' in 1 9" 1-1 5 to over 100/105 in 1970 (Doll and Peto, 19761. The mortality trends for the United States during the period 1950-75 are shown in Figure 2. The age-adjusted rates tor white males increased steadily from 24.6/105 during 1950-54 to 58.6/103 during 1970-75, with an even steeper incline among nonwhites. In recent years the rate or increase has been greater in females than in males, reflecting the growing popularity of cigarettes among females over the past 20 to 30 years. If present trends persist, in a few years lung cancer will become the leading cause of cancer death among women. The lung cancer trends for American men are 80 70 -- 566" ``` v :v.Lv. ^ v . Qo. Figure 2. Age-adjusted lung cancer mortality rates. 1950-75. in the United > States according to sex and race. rs.\ *i> O <i * ; r (------ l_______ j------------- ;________ i________ l 1950-54 1955-59 1960-64 1965-69 1970-75 Cj.encar Period ' *.' ;' ' j' ; '; *' ,` j ,I 1,' j] ' * i * J * i.1 H'j i MM * if illustrated by the age curves tor year-ot'-birih cohorts in Fiqure 3. Mortality increased almost linearly with age (on a log-log scale) except for some downward curvature in the oldest age groups, and was consistently higher in the later born cohorts. There was a remarkable shift in the rates of lung cancer according to race; for those born around 1385, the rates in whites exceeded those in nonwhites by about 50 per cent at all ages, while for those born around 1915 the situation was reversed and rates in nonwhites exceeded those in whites by about 50 per cent. The effects associated with more recent birth cohorts are diminishing, however, particularly for whites, with those born in the late 1920s and early 1930s as yet not showing rates appreciably different from those born around 1915-20. Lung cancer rates are also starting to level off in other countries. Cohort analyses in the Netherlands suggest that lung cancer mortality will peak among males born in the 1930s and then decline among those born thereafter (Van der Hoff, 1979). Lung cancer has recently decreased among British physicians, corresponding to a 60 per cent reduction in average cigarette consumption among this group over the past 20 years (Doll and Peto, 1976). It seems likely that the age-adjusted rates will reach a plateau in Western countries over the next few cfecades as those born in the 1920s and later constitute an increasingly larger share of the total population. Because of the fairly rapid reduction in risk following cessation of smoking, the rates mav eventually drop among population groups that follow the example of British doctors. Urbanization Lung cancer tends to be more common in urban than rural areas in virtually all parts of the world. Table 2 shows average lung cancer mortality rates for United States counties, 195069, according to urbanization level and geo graphic sector. Regardless of region of the country, mortality tended to be higher for both sexes in the more metropolitan counties. Although people in cities tend to smoke more than those on farms, a small urban excess is still evident after controlling for smoking habits (Public Health Service, 1979). l* 4f 1. I i1 I 1I ( /t jj t | { [ I I I | f | * t J: U c o. rv rv rv -- 567 l 20C- oz - 200- < 3 5 Figure 3. Age-soec.hc lung cancer 3C - * mortality rates. 1950-75 among white < 7o -- r N. ,t f and nonwhite mazes m me United States lor selected year-of-dirth co > 60 r50 - V horts h- ! < 40- \3 jt. 30 4 / < // / ij i'/ /' IS SO 55 SC 65 70 75 30 35 ^ AGE Social Class An inverse association between lung cancer and socioeconomic status has been observed in several studies A more than twofold difference in mortality between low and high social class, as measured primarily by occupation, is seen in recent British mortality data [Registrar General, 1 978) tFig. 4j. Similar results are seen in surveys measuring income or education (Graham et al, 1960; Seidman, 1970; Williams and Horm, 1 977). Smoking habits contribute to at least part of the socioeconomic differential ot lung cancer (Wynder and Stellman, 1977). TABLE 2. Average Annual Age-Ad|usted Lung Cancer Mortality Rates Among Whites Per 100.000 Population. 1950-69, for U.S. Counties According to Urbanization. Geographic Sector, and Sex' Sex Geographic Sector Males Females Northeast Southeast Midwest South Central North Central Mountain Far West Northeast Southeast Midwest South Central North Central Mountain Far West "From Blot and Fraumem (1976). 0-24 9 33 29 26 31 22 23 32 59 51 49 54 42 43 57 Urbanization (Per Cent) 25-49 9 50-74 9 35 32 29 34 24 24 33 56 51 50 53 4.7 46 62 38 39 33 36 27 27 34 5.6 60 54 59 48 51 62 75* 44 46 40 40 35 30 41 6.7 66 63 69 54 56 7.7 'll :i I W\'l o I U I 3CLJJ ICO - 568 -- 1. 1 \ t. - k ./N 1 ' K. v,, \ Figure 4. Standardized mortality ratios for lung cancer, according to social class, among males age 15-64. Great Britain, 1970-72 (from Registrar General, 1978). Religion In surveys in New York, Pittsburgh, and Mon treal. a 20 to 50 per cent reduction in risk was found among Jewish males, although some ex cess has been noted among Jewish females (Horowit.z and Enterline, 19/0). The deficit in males has been attributed mostly to lower ciga rette consumption, but the excess in females remains unexplained. In Israel, incidence among males is low bv western standards, but not among females whose tumors are mostly adenocarcinomas (Modan, 1978). At even lower risk are Mormons and Seventh-Day Ad ventists, whose religious dictates proscribe the use of tobacco products (Lyon et al, 1976; Phillips et al, 1980). Ethnic Groups 1974) and with the high incidence of lung adenocarcinoma reported in Chinese females in Singapore and Hong Kong (Chan et al, 1979; MacLennan et al. 1 977). Smoking habits do not explain this pattern. Histology Virtually all lung cancers are derived from epithelial tissue (McDowell et al, 1978). Al though there are several histologic types, squa mous cell (epidermoid) carcinoma is the most common form in western nations, predominates in males, and is closely linked to smoking habits. Next most common is adenocarcinoma, which constitutes a high percentage of cases in low-risk countries and m females, and is less closely related to smoking. A disproportionate Recent incidence data from the Surveillance, Epidemiology, and End Results (SEER) program of the .National Cancer Institute permit a break down for several ethnic and racial groups within the United States (Table 3). Low rates are seen for some groups, notably Hispanic males, American Indians, and Japanese, related in part to low tobacco consumption. The highest rates are seen in native Hawaiians and blacks. His panic American females were once reported to have increased rates for lung cancer, but recent data suggest that the excess no longer exists (Menck et al, 1975). The high rates among Chinese females are consistent with United States mortality data (Fraumeni and Mason, TABLE 3. Average Annual Age-Adjusted (1970 Standard) Lung Cancer Incidence. 1973-77. in SEER Registry Areas in the United States. According to Sex, Racial and Ethnic Groups Racial or Ethnic Group White Hispanic (New Mexico) Black Chinese (San Francisco Hawaii) Japanese (San Francisco, Hawaii) Hawaiian American Indian (New Mexico) incidence (Cases/YrnO'l Male Female 76 4 27 0 1100 62 3 49 1 116 9 10.8 21 8 17 7 24 3 32 0 13 6 50.7 3.9 1 ^LNC and PlELKA-- 569 recent increase in acenocarcinoma has been Great Britain supported the link (Doll and Hill, i noted but mav be related to changing diagnostic 1952; Levin et al. 1950; Wvnder and Graham, f practices iVmcent et al. I 9771. The third mo>t 1950k and the evidence became conclusive i common tvpe is small cell anaplastic carcino through survevs of large cohorts of smokers. The ma, which is more common in males and Public Health Service (1971) summarized 35 i closelv related to smoking. Other cell types, case-control and 8 cohort studies dealing with such as large cell carcinoma, alveolar cell car this issue. Every one, regardless of method or cinoma, and malignant carcinoid, are rela location, indicated an elevated relative risk (RR) tively uncommon. of lung cancer among smokers. Table 4 shows the RR according to amount smoked based on Survival data from the three largest cohort studies: the American Cancer Societv (ACS) follow-up over The National Cancer Institute has evaluated six years of 1 million persons (Hammond, the survival experience of patients with lung 1972); the 20-year assessment of mortality of cancer in the United States since 1950 (Axtell et 34,000 male British physicians (Doll and Peto, al, 1976). Although the traction of all cases 1976); and the 8 1 /2-year follow-up of 290,000 with localized disease has remained about 20 United States veterans (Kahn, 1966). All three per cent, the one-vear survival rate for white investigations revealed a steady rise in lung * patients increased from 20 per cent in 1950-54 cancer with increasing amount smoked, so that to 31 per cent in 1970-73, and the five-vear rate the risk for male smokers of two or more packs increased trom 6 to 9 per cent over the same per dav was nearly 20 times that of non- period. Females experienced better survival smokers. than males. Among patients with localized dis Similar trends were seen with other indices of ease, ihe rive-vear survival rates rose from 21 to cigarette consumption, including duration of 33 per cent. Improvements occurred also for smoking and total pack-vears of use. The ACS patients with regional disease, but were less survey showed a higher RR among males than pronounced. The corresponding survival pat females, a pattern seen in other studies as well. terns ror black patients were less favorable. The male excess is generally attributed to the Patients with small cell carcinoma, which had fact that women began smoking later in this the smallest proportion of localized tumors, century than men, started at older ages, and showed the poorest survival. inhaled less deeply (Hoover, 1978). The studies in the United States have shown an increased risk with inhalation, but the British data indicate Tobacco an inhalation effect limited to light or moderate smokers. Detailed investigations have uncov In the 1920s and 1930s clinical observations ered no confounding variable or bias that might and an alarming increase in lung cancer in affect the association between smoking and cidence raised some suspicion that smoking lung cancer, so that a cause and effect relation might be a causal factor. By the early 1950s ship is beyond question. It is noteworthy that case-control studies in the United States and tobacco smoke contains several agents, includ- TABLE 4. Relative Risks of Oeath From Lung Cancer According to Number of Cigarettes Smoked in Three Cohort Studies American Cancer Society (ACS) MALES FEMALES United States Veterans British Physicians Nonsmokers* 1.0 1.0 1.0 1 0 Current cigarette smokers'1 1-9 10-19 20-39 40 + 92 46 0.6 14 7 18.8 2.2 13 24 49 75 12.1 55 99 17.4 23 9 14 0 78 17.4 25.1 Did not smoke any form of tobacco. Classification refers to ACS study The categories for UnitedStates veterans were 1-9. 10-20, 21-39. and 40+, and for British physicians were 1-14. 15-24. and 25+ cigarettes per day I I' iil. i v" c c il c rx fN o '1 u l \ ![!.: k * iUtH#-r * Or' ` kW; tv \W ip ' I ! l! i: li i 'll .u ii QP ? 7 F I n K 570 -- Cancel 5s mg poKc.c <; hvnro jaroons that are carcin Lnitec; States durirg 'lu58 Haenszei et a., ogenic m Uooratorv animals (Wvnder ana Hotf- 1 9621 mann. I'Vfji It is estimated that m the United States ciga The risk oi lung cancer is related to the t\ pe or rette smoking mav contribute to at least 80 per tobacco product smoked. Although the effect or cent of lung cancer in males and 40 per cent in smokmc low tar and nicotine cigarettes is not females iHammond and Seidman, 1980). Smok set ssell quantiiied. case-control studies have ing will continue to be the dominant risk factor shown some reduction in risk tor users of filter tor lung cancer in the sears to come, but its compared to nontilter cigarettes even after con relative importance may decline if recent trends trolling tor the amount smoked (Wynder and toward reduced cigarette consumption and Stellman, 1977). Recent data from the ACS lower tar and nicotine yields are maintained survey hase also shown some lowering in mor (Wynder and Hoffmann, 1979). More remains tality, bv about 20 per cent, tor male users or to be learned about the changing risk of various losv compared to high tar and nicotine ciga diseases following alterations in smoking habits rettes (Hammond et al, 1977). The greatest and product modification, and the extent to reduction in risk comes from cessation of smok which smoking interacts with other environ ing (Table 5), as several studies have shown a mental and host determinants of lung cancer. substantial lowering of mortality among long term ex-smokers (Public Health Service, 1979). Increased risks of lung cancer have also been Occupational Factors associated with pipe and cigar smoking, but these are of a lower order of magnitude than the Although cigarette smoking is the principal risks seen with either filter or nonniter cigarettes cause ot lung cancer, studies of occupational <e g , RR -- 2 in the ACS and United States groups have led to the discoserv of several seterans studies and RR - 7 among British respiratory carcinogens iFraumem, 1975). pnssiciansi It has been suggested that cigarette There are probably other hazards that remain to I; smoke is less irritating and thus more easily be identified in the workplace, so it is difficult at ji .nhaled than pipe or cigar smoke. present to estimate the oserall contribution of Smoking seems to induce lung cancers of all occupational factors to cancers of the lung and the maior histologic types (Vincent et al, 1977; other sites. It is also significant that the effects of Wvnaer and Stellman, 1977). The strongest some occupational carcinogens (asbestos, associations are for squamous cell and small radon) are greatly enhanced by tobacco cell carcinomas, but dose-response relation smoke. ships for adenocarcinomas and other cell types have also been reported. Figure 5 shows in Asbestos creasing mortality ratios with amount smoked for squamous cell or small cell (undifferentiat Over the past 25 years epidemiologic evi ed) carcinomas and tor adenocarcinoma. The dence has accumulated indicating that the risk data were obtained from family informants and of lung cancer, mesothelioma, and asbestosis is physicians tor 2,381 lung cancer deaths, repre substantially raised among workers in various senting a 10 per cent sample of all deaths In the asbestos industries, including miners and mill- TABLE 5. Relative Risks* of Death From Lung Cancer According to Years Since Cessation of Cigarette Smoking British physicians' American Cancer Society 1-19 cigarettes/day 20-*- cigarettes/day Years Since Stopped Smoking 0 <5 5-9 15 8 10.7" 59 7 1 3.3 1 3 17 1 10.1 65 10-14 53 7 5-i2.0 0.4 1.8 All risks relative to nonsmokers of any form of tobacco. All individuals smoked cigarettes at least 5 years and began before age 25 These who stopped had smoked about 10 per cent fewer cigarettes per day than those of the same age who continued. "Excluded men who quit smoking after the diagnosis of lung cancer. STO!32231 \C J\D -- 571 10?0000 800 700 600 500 400 300 200 Figure 5. Standardized mortality ratios for squamous cell,undifferentiated carcino mas and adenocarcinoma of the lung by smoking category among white males in the United States. 1958 (based on data from Haenszel et al. 1962) 100 cn 90 80 70 CO 60 50 40 30 20 10 9 8 6 5 EcMdermoid and Undifferentiated Carcinoma 11296 Deathsl Adenocarcinoma (325 Deathsl IV ty Smoking Categories: l-Never Smoked II-Ex-cigarette Smoker III-Regular Cigarette Smoker Kl Pack/Day) IV-Regular Cigarette Smoker i>1 Pack/Dayl ers, and textile, insulation, shipyard, and ce ment '.corkers (Lemen et al, 1980). Lung cancer is the major asbestos-related disease, resulting m apout 20 per cent of all deaths in some exposed cohorts, with higher percentages in subgroups with greater intensity and duration of exposure (Newhouse and Berry, 1979; Selikotf et al, 1979). The length of exposure is not necessarily long, however, as indicated by the doubling of risk 20 years after working tor less than nine months in a United States amosite factory (Seidman et al, 1 979), and by the 60 to 70 per cent excess risk that now exists among American men who worked temporarily in ship yards during World War II (Blot et al, 1978, 1980). Because of its carcinogenic potential and widespread distribution m many industries, as bestos is the agent generally considered to pose the largest carcinogenic threat in the work place. Although exposures are usually to mixed forms of asbestos, all fiber types (chrysotile, amosite, crocidolite, anthophyllite) are thought to increase the risk of lung cancer (Lemen et al, 1980). The asbestos-related tumors tend to arise more often in the lower lobes, but all histologic types of lung cancer seem to be affected. In most studies the risk of asbestos-induced lung cancer is characterized by a latent period of 20 years or longer between start of exposure and onset of the disease. In a mortality follow up of 17,800 United States and Canadian asbes tos insulation workers, the excess of lung cancer was twofold during the period of 10 to 14 years after initial employment, reached nearly sixfold between 30 to 34 years after employment, and declined at longer intervals (Selikotf et al, 1979). Another characteristic of asbestosinduced lung cancer is its synergistic relation with cigarette smoking (Saracci, 1 977). A recent survey of 276 lung cancer deaths among insula tion workers (Hammond et al, 1979) revealed relative risks of about 5 following asbestos exposure alone, 10 following cigarette smoking alone, and 50 following combined exposure (Table 6). Calculation of mortality ratios for this cohort according to the amount smoked showed a gradient in risk, with heavilv smoking asbestos workers experiencing nearly 90 times the risk of individuals who were not exposed to tobacco or asbestos. Similar results have been seen among amosite factory employees (Selikotf TABLE 6. Relative Risks of Lung Cancer According to Asbestos Exposure and Cigarette Smoking Status' Asbestos No Yes Cigarette Smoking No Yes 1 0 10 9 5.2 53.2 'From Hammond et al (1979).