Tobacco Documents Online

EASE :onal alcohol 's and hnpli- Blanchet, F., K, P~quinot, I Zubiri, L., ntemational ~rra (Spain), ~-491,1988. • D. A., and m Brazil: a ~d Terracini, Jder cancer, and Roy, P., ,unbeds and ~son, G. M., a myeloma: nder, B. G., :o phenoxy m, J. Natl. 10 LUNG CANCER G6ran Pershagen CONTENTS Introduction ................................................. 147 Lung Cancer Occurrence ...................................... 148 Environmental Causes of Lung Cancer ........................... 148 Tobacco Smoking and Occupation ............................ 148 Ambient Air Pollution ...................................... 149 Urban Areas ............................................. 149 Industrial Areas ......................................... 150 Residential Radon ......................................... 151 Cooking Fumes ........................................... 152 Environmental Tobacco Smoke ............................... 153 Research Needs .............................................. 154 Role of Vital Statistics ...................................... 154 Definition of Disease ...... . ................................. 154 Individual Susceptibility .................................... 155 Markers of Exposure ....................................... 155 Hypotheses to Be Tested . : .................................... 156 References .................................................. 157 INTRODUCTION Lung cancer is believed to be the most common fatal neoplastic disease in the world today (IARC, 1986). In comparison with other types of cancer, many etiological factors have been identified, which together often account for a major part of the cases. Tobacco smoking is the dominating cause and occupational exposures are of importance in some situations. However, for many environmental exposures, which may be of great significance from a public health point of view, the evidence remains unclear. 147

148 ENVIRONMENTAL EPIDEMIOLOGY: EXPOSURE AND DISEASE LUNG CANCER This review will focus on lung cancer in relation to environmental expo- sures such as ambient air pollution, residential radon, environmental tobacco smoke (ETS), and combustion products from indoor sources. Tobacco smoking and occupational exposures are taken up mainly as interacting and confounding factors and for risk extrapolation from high exposures. Identification of research needs will receive particular attention, especially in relation to the application of new molecular biologic techniques in epidemiologic studies. LUNG CANCER OCCURRENCE The incidence of lung cancer has risen sharply during this century, and it is now one of the leading causes of cancer death in many countries. For example, in the U.S. the age-adjusted lung cancer mortality in men increased from 11 per 100,000 in 1940 to 73 per 100,000 in 1982 (Garfinkel & Silver- berg, 1990). The lung cancer death rate in women started to rise in the early 1960s from 6 per I00,000 to 25 per 100,000 in 19~6. In the younger age groups there has been a levelling off in the mortality rates among men but not in women. In Sweden the lung cancer incidence has increased an average of 1.7 and 3.9% yearly in men and women, respectively, during the last two decades (Pershagen, 1990). Lung cancer rates among middle-aged men in Eastern Europe are approaching the high rates in Finland and the U.K. of the 1950s (IARC, 1986). Also in many developing countries there is a sharp increase in lung cancer incidence and by the turn of the century it is estimated that around 2 million lung cancer deaths will appear annually in the world. There are pronounced differences in lung cancer incidence between dif- ferent countries and regions. The highest age standardised incidence rates in an international comparison were found in black men of New Orleans, U.S. (110 per 100,000 and year) and in Maori women of New Zealand (68 per 100,000 and year). The lowest rates in men and women were found in Madras, India (6 and 2 per I00,000 and year, respectively). In general, urban areas showed higher rates in both men and women than rural areas (IARC, 1987). The major differences in lung cancer occurrence between high and low incidence areas indicate that there is a substantial potential for prevention. Furthermore, in spite o'f great efforts to improve treatment, most lung cancer patients die within i year after diagnosis. Primary prevention must thus be given high priority. ENVIRONMENTAL CAUSES OF LUNG CANCER Tobacco Smoking and Occupation Tobacco smoking is the most important cause of lung cancer, and where prolonged smoking is widespread it generally accounts for more than 80% of the cases (IAR( of smoking. Aft falls and will a1 Peto, 1976). Stl lung cancer in Rogot & Murra 1987; Carstensc makes it impov in all studies ot A number, increased risk c polyaromatic h, vidual risks ma cancer due to o number of subj tionally expose proportion ma) In several between smoki~ pies include as~ and radon daug esis with agent may generate tt less pronounce Radford & St. conclusively re. of a small nul statistical powe Ambient Air Urban Areas The comF complex. Som~ demiologic stu 1950s; and urb to the 1980s. t for residential while in other,, "urban" thus d differences, be lutants and the

LUNG CANCER 149 tl expo- tobacco moking ~unding .tion of ~ to the udies. , and it es. For creased Silver- ~e early zer age but not rage of tst two ~nen in • of the .'orld. en dif- ates in s, U.S. 68 per !adras, i areas 1987). ]d low ration. cancer .ms be where 0% of the cases (IARC 1986). The risk is closely related to duration and intensity of smoking. After cessation of smoking the excess relative risk of lung cancer fails and will approach zero after 10-20 years (Cederl6f et al., 1975; Doll & Peto, 1976). Studies from the U.K. and the U.S. have shown lower risks for lung cancer in pipe smokers than in cigarette smokers (Doll & Peto, 1976, Rogot & Murray, 1980), but this is not true for Sweden (Damber & Larsson, 1987; Carstensen et al., 1987). The dominating role of smoking in lung cancer makes it important to be considered as an effect modifier and/or confounder in all studies of lung cancer etiology. A number of occupational exposure factors have been associated with an increased risk of lung cancer, such as asbestos, chromates, inorganic arsenic, polyaromatic hydrocarbons, and radon (Simonato et al., 1988). Although indi- vidual risks may be substantial, the population attributable proportion of lung cancer due to occupational exposures is generally low because of the limited number of subjects exposed. However, in some situations where the occupa- tionally exposed constitute a large fraction of the population, the attributable proportion may reach 40%. In several studies on lung cancer etiology, a multiplicative interaction between smoking and occupational exposures has been observed. Some exam- ples include asbestos (Hammond et al., 1979), arsenic (Pershagen et al., 198 I), and radon daughters (Archer et al., 1973). Multistage models for carcinogen- esis with agents operating at different stages in the cancer induction process may generate this type of interaction. In other studies, the interaction has been less pronounced, more consistent with an additive effect (Pinto et al., 1978, Radford & St. Clair-Renard, 1984). In most studies it was not possible to conclusively reject either an additive or a multiplicative model, mainly because of a small number of non-smoking lung cancer cases resulting in a low statistical power of the test. Ambient Air Pollution Urban Areas The composition of ambient air in urban areas is quite variable and complex. Some examples of the environments under investigation in the epi- demiologic studies reviewed here include British towns and cities during the 1950s; and urban areas in Japan, China, Europe, and the U.S. from the 1960s to the 1980s. Emissions resulting from the use of coal and other fossil fuels for residential heating were dominating sotffces of pollution in some areas, while in others motor vehicles or industries were more important. The term "urban" thus denotes a mixture of environments, which may show substantial differences, both in terms of actual exposures to various environmental pol- lutants and the influence of interacting or confounding factors.

150 ENVIRONMENTAL EPIDEMIOLOGY: EXPOSURE AND DISEASE LUNG CAI~ The epidemiologic evidence on air pollution and lung cancer has recently been reviewed by Pershagen and Simonato (1990) and will be discussed only briefly here. Seven cohort studies on urban air pollution and lung cancer were available. All but one of the investigations contained information on smoking for all study subjects. The studies came from the U.S. (3), Sweden (2), Finland (1), and the U.K. (1). Smoking-adjusted relative risks for lung cancer in urban areas were generally on the order of 1.5 or lower in those cohort studies reporting increased risks. The findings pertain mainly to smokers. For non- smokers the number of cases was generally too small for a meaningful inter- pretation of urban-rural differences. In the 13 case-control studies reviewed by Pershagen and Simonato (1990), residential and smoking histories were obtained for the study subjects and sometimes also information on potential confounding factors, such as occupation. Increased relative risks for lung cancer were observed among men in urban areas in three British studies as well as in studies from Greece, Poland, China, and Japan. Two U.S. studies found raised lung cancer risks in urban males, while another two failed to show an effect. The results for women are difficult to interpret because of small numbers, but at least one study indicated a raised lung cancer risk for females in urban areas, also among nonsmokers. The magnitude of the excess relative risks for lung cancer in urban areas reported in the case-control studies was similar to that in the cohort studies. The epidemiologic studies on urban air pollution and lung cancer gave somewhat inconsistent results as to the type of interaction with tobacco smok- ing. Some studies provided evidence of a combined effect exceeding an addi- tive effect, and often compatible with a mtdtiplicative interaction, while other studies were more consistent with an additive effect. Industrial Areas Several epidemiologic studies have been carried out in areas near copper, lead, or zinc smelters (Pershagen & Simonato, 1990). The emissions from the smelters are quite complex, but inorganic arsenic is often a major component. The studies come from four countries (Canada, China, Sweden, and U.S.) and are of ecologic or case-control design. Five ecologic studies showed increased lung cancer rates,among men living in areas near non-ferrous smelters with relative risks ranging from about 1.2 to over 2. Two case-control studies showed relative risks of 1.6 and 2.0 for men living near the smelters after adjustment for occupation and smoking. A study of women near a U.S. smelter also suggested an increased lung cancer risk related to estimated exposure to arsenic in ambient air. Ecologic studies on lung cancer have been performed in areas with indus- tries of different types, including chemical, pesticide, petroleum, shipbuilding, steel, and transportation industries (Pershagen & Simonato, 1990). Most of the studies showed increased lung cancer risks, which did not seem to be explained trolled or It is ~ lution ant this relati, informati~ making it dose-resp uted to th Residen ~ nu exposed t (NAS, 19 daughters in many 1 an impor~ for severe data in m Lung with diffe case-cont acteristic~, between e sizes and data for c Five measuren residency Ruosteen detailed : Positive ~ ies, while in expost comparis, The • cancer m based on 1 1987) or estimates Ruosteen, range as suggests ~

LUNG CANCER 151 :ently [ only were oking nland urban udies non- inter- explained by socioeconomic factors. However, smoking habits were not con- trolled or was employment at the industries under study. It is difficult to interpret the epidemiologic evidence on ambient air pol- lution and lung cancer. Many studies were not originally designed to study this relation which has implications for the detail and quality of the exposure information. Data from measurements of air pollutants were generally limited, making it difficult to compare the findings in different studies and to assess dose-response relationships. Uncontrolled confounding may also have contrib- uted to the results. 3nato ~jects :h as : men ,land, ~rban n are cated ,kers. areas lies. pper, ,~ the ilent. ~ and ased with tdies after elter re to dus- ling, it of ~ be Residential Radon A number of epidemiologic investigations show that underground miners exposed to high levels of radon daughters run an increased risk of lung cancer (NAS, 1988). Studies in experimental animals confirm that inhalation of radon daughters can induce lung cancer. The increased radon concentrations found in many homes of some countries, suggest that residential radon exposure is an important risk factor for lung cancer in the general population. However, for several reasons, quantitative assessments of population risks based on the data in miners and experimental animals are uncertain. Lung cancer risks related to residential radon exposure have been studied with different epidemiologic methodologies (Samet, 1989). Earlier cohort and case-control studies based the exposure estimation primarily on housing char- acteristics and/or geology. Many of these studies showed an association between estimated radon exposure and lung cancer risk; however, limited study sizes and imprecision in the exposure estimates make it difficult to use the data for quantitative risk assessments. Five recent case-control studies based the exposure estimation on radon measurements in homes of the study subjects covering about 10-30 years of residency (Axelson et al., 1988; Blot et al., 1990; Schoenberg et al., 1990; Ruosteenoja, 1991; Pershagen et al., 1992). These studies also contained detailed information on smoking and other potential confounding factors. Positive exposure-response relationships were suggested in some of the stud- ies, while others found no evidence of a trend. Differences between the studies in exposure levels and in the influence of other risk factors complicate a comparison of the results. The risk estimates in the studies on residential radon exposure and lung cancer may be compared with those obtained from miners. Risk estimations based on the mining data often used relative risk models, either constant (ICRP, 1987) or modified by age and time since exposure (NAS, 1988). The risk estimates from the residential radon studies by Schoenberg et al. (1990), Ruosteenoja (1991), and Pershagen et al. (1992) appear to lie within the same range as those projected from miners, while the study by Blot et al. (1990) suggests a lower risk. There are considerable uncertainties in the comparisons

152 ENVIRONMENTAL EPIDEMIOLOGY: EXPOSURE AND DISEASE LUNG with the mining data. For example, age differences in risk or consequences of a nonmultiplicative interaction between radon and smoking have not. been considered. The type of interaction between radon exposure and smoking is of impor- tance for the risk assessment. In miners a multiplicative or submultiplicative interaction was often found but the data are not fully consistent (NAS, 1988). A major difficulty in the interpretation arises from the fact that the number of lung cancers is low among nonsmokers, although the data clearly indicate that the risk is elevated also in this group among radon exposed miners. For residential exposure only limited evidence is available and no clear pattern of interaction between smoking and radon exposure has emerged. Cooking Fumes Several epidemiologic studies have investigated lung cancer risks in rela- tion to exposure to combustion products in homes. The sources include fuels used for cooking and heating, such as coal and kerosene, as well as vegetable oils used for frying, etc. Most of the studies come from China, where indoor air pollution levels from such sources may be substantial (Mumford et al., 1987). In this context it is worth noting that increased lung cancer risks have been reported both among cooks and bakers (Coggon et al., 1986; T~ichsen & Nordholm, 1986), and among coke oven workers (Redmond, 1976). Two case-control studies on female lung cancer from China suggest that cooking oil fumes may be of importance for lung cancer (Gao et al., 1987; Wu-Williams et al., 1990). In one of the studies there was an increased risk of lung cancer associated with the use of rapeseed oil for cooking compared to soybean oil. The lung cancer risk was related to degree of smokiness when cooking and number of dishes prepared per week. In the other study increased risks of lung cancer were related to number of meals prepared by deep frying and to eye irritation during cooking. In parts of China indoor burning of coal for domestic cooking and heating is common. Two types of coal are "smoky coal", which smokes heavily on firing; and "smokeless coal", which produces little smoke. Lung cancer rates in Xuan Wei county among both men and women living .in areas where "smoky" coal was the predominant fuel were substantially higher than in areas where "smoke'~ess" coal or wood was used (Mumford et al., 1987). A subse- quent case-control study indicated that domestic coal use was a stronger risk factor for lung cancer in females than smoking and that the risk was related to the duration of exposure (He et al., 1991). In the study by Wu-Williams et al. (1990) there was an elevated risk related to the number of years using coal for domestic heating. A case-control study from Guangzhou, China showed that several vari- ables indicating a poor kitchen and house ventilation were associated with an increased lung cancer risk (Liu et al., 1992). Coal was the predominant fuel used i burnin al., 19 A increa: risk ,s petrok lung c~ et al., It for car where the do oils us Envin Er (SS) re exhale~ of mai emittec more ti side st1 ETS is the int~ Ex its met measur pariso~ cotinin afewi Pursiar reporte ETS (( Th Persha~ 3000 n. gesting observz cohort : Greece, (more t

LUNG CANCER 153 ences of ~ot been .f impor- plicative ;, 1988). tmber of cute that .ers. For attern of ~ in rela- tde fuels egetable 'e indoor :d et al., sks have .chsen & :7; ~sed risk ompared ~'ss when ncreased frying heating ,avily on cer rates ts where in areas A subse- nger risk related lliams et qing coal :ral vari- t with an nant fuel used in this area. An elevated relative risk associated with residential coal burning during childhood was observed in a U.S. case-control study (Wu et al., 1985). A case-control study from Hong Kong (Koo et al., 1983) indicated an increased lung cancer risk in women associated with kerosene use. No excess risk was observed for other types of fuels, such as wood/grass or liquid petroleum gas. However, in a study of women from Osaka, Japan, an increased lung cancer risk was related to the use of straw or wood as cooking fuel (Sobue et al., 1990). It may be concluded that indoor combustion sources can be of importance for cancer of the respiratory tract. In some situations, such as in parts of China where smoking is rare and indoor burning of coal is common, it seems to be the dominant cause of lung cancer. It is possible that fumes from vegetable oils used for cooking may also play a role, but these data need confirmation. Environmental Tobacco Smoke Environmental tobacco smoke is produced from the side stream smoke (SS) released from the burning tobacco product and mainstream smoke (MS) exhaled by the smoker. There are substantial differences in the composition of mainstream and side stream smoke (NRC, 1986). As a rule the amounts emitted from a cigarette are greater in the side stream smoke. For example, more than 50 times higher amounts are emitted of some nitrosamines in the side stream smoke, but the SS/MS ratio is around 2-5 for most components. ETS is predominantly an indoor problem; and the exposure level depends on the intensity of smoking, room size, and air exchange. Exposure to ETS can be estimated with biologic markers. Nicotine and its metabolite cotinine are two sensitive and specific markers which may be measured in saliva, serum, and urine. Cotinine has some advantages in com- parison with nicotine, including a longer biologic half-life. The levels of cotinine in nonsmokers reporting ETS exposure range from about one half to a few per cent of those in regular smokers (Jarvis et al., 1984; Husgafvel- Pursianen et al., 1987). High urinary cotinine concentrations have been reported in children with smoking parents indicating a substantial exposure to ETS (Greenberg et al., 1984; Henderson et al., 1989; Rylander et al., 1989). The evidence on ETS and lung cancer has recently been reviewed by Pershagen (1994) and will only be discussed briefly here. A total of almost 3000 non-smoking lung cancer cases have been included in 27 studies, sug- gesting a total study base on the order of 40 million person-years under observation. Most of the studies were of case-control design, and only 3 were cohort studies. The studies come from six countries (China with Hong Kong, Greece, Japan, Sweden, U.K., and U.S.) and include predominantly women (more than 90% of the cases).

154 ENVIRONMENTAL EPIDEMIOLOGY: EXPOSURE AND DISEASE Exposure to ETS was defined in different ways in the studies; however, very often the exposure classification was based on smoking habits of spouses. Combining the results of the studies gives pooled relative risks of lung cancer in nonsmoking women and men living with smokers of 1.23 Ii.11-1.36) and 1.82 (0.98-3.37), respectively. It is necessary to consider various sources of bias and their implications in the assessment of the epidemiologic evidence on passive smoking and lung cancer. This is particularly relevant in view of the weak associations observed. Probably the most important source of bias is confounding by unreported active smoking. There is a tendency for smokers to man3.' smokers; and in conjunction with some misclassification of smokers as nonsmokers, this would tend to produce a spurious association bet~veen spouse smoking and lung cancer. Unfortunately there is limited empirical information which can be used to estimate the extent of this bias, and differences in assumptions may lead to quite discrepant conclusions regarding its magnitude (Wald et al., 1986; Lee, 1988). Other types of bias, such as non differential misclassification of ETS exposure and lung cancer, may give rise to an underestimation of the true risk. RESEARCH NEEDS Role of Vital Statistics As indicated previously cancer and mortality registers have played a significant role in studies of lung cancer occurrence and etiology. Because of the dominating role of tobacco smoking for lung cancer, descriptive data on geographical differences and time trends will mainly reflect earlier and present smoking habits. However, there are many examples where descriptive infor- mation on lung cancer occurrence has led to the identification of occupational and environmental hazards (Pershag~n & Simonato, 1990). It would be par- ticularly useful to obtain more data on lung cancer occurrence in areas with heavy air pollution, such as in parts of Central and Eastern Europe. The high lethality of lung cancer implies that information on its occurrence also may be obtained also using mortality data. Furthermore. it is possible to assess the quality of both mortality and cancer morbidity registers by com- paring the da,~a on lung cancer. For many reasons it is of great importance to have high quality registries on mortality and cancer morbidity. Besides pro- viding useful descriptive information, they may also serve as a source of cases in analytical studies. Thus, efforts should be made to support existing registries and to create new registries, as well as to continuously assess their quality. International collaboration is vital in these activities. Definition of Disease The term lung cancer mainly denotes primary cancer of the bronchus or lung. Generally, the quality of this diagnosis is high. However. in nonsmokers, whe lun[ one (Gal 198" cam the ~ lung inte~ ronn exar cino type niun is a sure reso gene supp cers foun two al., 1 for 1 Indi sugg a po: meta ity do v, for f bias popu effec Marl ( whic

LUNG CANCER 155 however, f spouses. ng cancer 1.36) and plications • and lung observed. nreported "s; and in his would and lung n be used ty lead to 386; Lee, n of ETS true risk. a .'cause of ; data on d present ve infor- apational t be par- • eas with currence ,ssible to by com- rtance to des pro- of cases 'egistries quality. • ~chus or ,mokers, where this tumour is rare, substantial misclassification may occur. Secondary lung tumours and carcinomas with unknown primary site appeared in about one sixth of reported cases of lung cancer on death certificates in the U.S. (Garfinkel, 1981) and in central health registers in Sweden (Pershagen et al., 1987) among female nonsmokers. Descriptive data on time trends of lung cancer in non-smokers must be interpreted with caution and the validity of the diagnostic information should be carefully assessed in analytic studies of lung cancer in nonsmokers. Lung carcinomas can be separated into different histological types, and international criteria of classification have been proposed (WHO, 1982). Envi- ronmental exposures may be associated with specific histological types. For example, tobacco smoking primarily induces epidermoid and small cell car- cinomas, although increased relative risks are also found for other histological types (IARC, 1986). Small cell carcinomas seem to predominate among ura- nium miners, especially in younger age groups (NAS, 1988). However, there is a great need for further data on associations between environmental expo- sures and specific histological types of lung cancer, which may increase the resolution power of the studies. New developments in molecular biology have led to the identification of genes which are involved in cancer induction, such as oncogenes and turnout- suppressor genes. Ki-ras and p53 mutations are common in human lung can- cers (Bos, 1990; Hollstein et al., 1991). A recent study of uranium miners found differences from the usual lung cancer mutational spectrum in these two genes, which may reflect the genotoxic effects of radon (V~ih~ikangas et al., 1992). Further studies should be performed to identify relevant mutations for lung cancer caused by environmental exposures. Individual Susceptibility A genetically determined individual susceptibility to lung cancer has been suggested. Variation in the ability to metabolise xenobiotics is considered as a possible explanation. Interest has focused on polymorphism of debrisoquine metabolism, aryl hydrocarbon hydroxylase, and glutathione transferase activ- ity (Idle, 1991). The evidence is not entirely consistent, which may have to do with differences in design and selection of study subjects. There is a need for further studies applying modern epidemiologic methodology to control bias in the assessment of individual susceptibility to lung cancer. If high risk populations could be identified, there would be better possibilities of detecting effects of environmental exposures. Markers of Exposure Crude exposure measures is a major problem in analytic epidemiology, which will generally lead to a dilution of any associations if the imprecision

156 ENVIRONMENTAL EPIDEMIOLOGY: EXPOSURE AND DISEASE is unrelated to the disease under study (nondifferential misclassification). New developments in molecular biology have created possibilities for increasing the precision of the exposure measurements. For example, exposure to geno- toxic agents may be monitored by measurement of DNA-adducts and muta- tions of marker genes, such as the hypoxanthine-guanine-phosphoribosyl- transferase (hprt) gene. Increased levels of DNA-adducts and hprt-mutations have been observed following exposures both in the occupational and general environment (Hemminki et al., 1990a; Hemminki et al., 1990b; Bridges et al., 1991; Tates et al., 1991). Biologic markers of long-term exposures extending over years or decades would be particularly useful for epidemiologic studies. HYPOTHESES TO BE TESTED There is a need fo~ further studies on the effects of ambient air pollution by chemical type or species on lung cancer. In particular, very few studies are available on lung cancer risks in areas where motor vehicles constitute the dominating source of pollution. Studies using modem epidemiologic method- ology should also be conducted in heavily polluted areas, such as in parts of Central and Eastern Europe, to generate data'for risk identification under different exposure situations. Current risk estimates based on miners suggest that residential radon exposu.re is the second most important cause of lung cancer in some countries and regions. However, more data are needed from epidemiologic studies on radon exposure in residences. For the purpose of increasing the present uncer- tainty in the risk estimation, these studies should be sufficiently large and involve radon measurements of residential periods covering several decades. More than 25 epidemiologic studies have been published on passive smoking and lung cancer. Taking the studies together, there is a statistically significant increase in the lung cancer risk of about 20-30% in nonsmokers married to smokers. It is probable that a part of this increase is explained by confounding by smoking. More empirical information on the degree of under- reporting of smoking by nonsmokers is needed to estimate the influence of this type of bias. A few epidemiologic studies indicate that indoor combustion sources may be of importance for lung cancer. Since this type of exposure is common in some parts ~f the world, particularly for women, there is a need for further evidence as a basis for the risk assessment. New developments in molecular biology should be utilised to a greater extent in environmental epidemiology. For example, such techniques may be used to identify sensitive groups, to increase the precision of the exposure measures, and to enhance the specificity of the effect determinations. The employment of molecular biologic techniques in epidemiology could increase the resolution power of the studies and contribute to elucidating mechanisms of toxic effects. Herr