Outdoor Air Pollution and Lung Cancer

Aaron J. Cohen
Health Effects Institute, Cambridge, Massachusetts, USA

In the 1950s evidence of an ongoing epidemic of lung cancer in the United States and Western The data collected by the U.S. EPA on
Europe led researchers to examine the role of outdoor air pollution, which was considered by some the criteria pollutants reflect, for the most
to be a likely cause. Although epidemiologic research quickly identified the central role of cigarette part, outdoor air pollution over relatively
smoking in this epidemic, and despite progress in reducing outdoor air pollution in Western large geographic areas. However, the exposure
industrialized countries, concerns that ambient air pollution is causing lung cancer have persisted to of human populations to carcinogens in out-
the present day. This concern is based on the fact that known carcinogens continue to be released door air may be the result of proximity to
into outdoor air from industrial sources, power plants, and motor vehicles, and on a body of more localized sources such as industrial facil-
epidemiologic research that provides some evidence for an association between outdoor air ities, small businesses (e.g., automotive body
pollution and lung cancer. This article reviews the epidemiologic evidence for this association and or chrome-plating shops), municipal facilities
discusses the limitations of current studies for estimating the lung cancer risk in the general (e.g., waste incinerators), or areas with high
population. It also identifies research needs and suggests possible approaches to addressing vehicular traffic. For example, data collected
outstanding questions. Key words: air pollution, ambient air, epidemiology, fossil fuels, lung cancer, by Cass and colleagues (9) in Los Angeles,
polycyclic aromatic hydrocarbons, review. - Environ Health Perspect 1 08(suppl 4):743-750 (2000).
http.//ehpnetl.niehs.nih.gov/docs/2000/suppl-4/743-750cohen/abstract.html California, indicate that elemental carbon lev-
els, mostly derived from diesel exhaust,
declined in five of seven areas in Los Angeles
between 1958 and 1981 but increased in two
In the 1950s evidence of an ongoing epi- sources such as building materials and vehicular areas undergoing rapid growth.
demic of lung cancer in the United States brake linings. Under the Clean Air Act Amendments of
and Western Europe led researchers to exam- Unfortunately, there are few long-term 1990, the U.S. EPA is charged with evaluating
ine the role of outdoor air pollution, which trend data for outdoor levels of known car- the risks of 189 hazardous air pollutants, pri-
was considered by some to be a likely cause. cinogenic products of fossil fuel combustion marily carcinogens from point sources. The
Although epidemiologic research quickly that could be used to estimate long-term combustion of fossil fuels for transportation
identified the central role of cigarette smok- exposures for epidemiologic purposes. and power generation contributes to the pres-
ing in this epidemic (1,2), a body of epi- Available data indicate that over the past ence of these many known or suspected car-
demiologic research over the ensuing 40 20-30 years improvements have been made cinogens in outdoor air. Some of the
years provides some evidence for an associa- in some indices of air quality. According to a potentially more significant pollutants in terms
tion between air pollution and lung cancer. report of the Council for Environmental of exposure prevalence and/or lung carcino-
This article reviews this evidence and dis- Quality (6), levels of benzo[a]pyrene in urban genicity are discussed in the following sections:
cusses outstanding questions and research air decreased 70% between 1970 and 1980.
needs. Recent monographs and books pro- Daisey et al. (7 reported that levels of sulfates Polycyclic Organic Matter
vide more comprehensive reviews of the liter- and particulate-associated organic matter Polycyclic organic matter (POM) as defined
ature on air pollution from both indoor and declined by 30-40% between 1964 and 1983 by the U.S. EPA in the Federal Clean Air Act
outdoor sources and the occurrence of lung in two industrialized New Jersey locations in in 1971 comprises a large and varied class of
cancer (3-5). the United States. chemical compounds, including polycyclic
The U.S. Environmental Protection aromatic hydrocarbons (PAHs) and nitro-
Exposures to Carcinogens Agency (U.S. EPA) collects data on six pollu- PAHs, which are known carcinogens and
in Outdoor Air tants for which the U.S. government has pro- mutagens (10) and are found in both the par-
Despite progress in reducing outdoor air pollu- mulgated national air quality standards, the ticulate and gas phases of outdoor air. In addi-
tion in Western industrialized countries, car- "criteria pollutants." Particulate matter with tion to those compounds released directly into
cinogens continue to be released into outdoor an aerodynamic diameter < 10 lpm (PM o) is the environment by combustion processes,
air from industrial sources, power plants, and the criteria pollutant of greatest current others are created from primary combustion
motor vehicles. Outdoor air, particularly in interest with respect to lung cancer because products, such as those emitted by diesel
densely populated urban environments, con- particles < 10 pm can be inhaled into the engines, via chemical and photochemical
tains a variety of known human carcinogens lung and carry carcinogenic substances on
(Table 1). These substances are present as com- their surfaces. Comprehensive monitoring of
ponents of complex mixtures that may include PM1o has only been in place since 1988; This article is part of the monograph on Environmental
carbon-based particles to which the organic before 1988 the U.S. EPA monitored only and Occupational Lung Diseases.
compounds are adsorbed, oxidants such as total suspended particulates (TSP), which Address correspondence to A.J. Cohen, Health
Effects Institute, 955 Massachusetts Ave., Suite 700,
ozone and sulfuric acid in aerosol form. The include particles too large to be inhaled into Cambridge, MA 02139 USA. Telephone: (617) 876-6700.
combustion of fossil fuels for power generation the lung. Decreases in TSP were observed Fax: (617) 876-6709. E-mail: [email protected]
or transportation is the source of most of the over the 1970s and early 1980s, but little Parts of this review appeared in earlier reviews by
the author, specifically Cohen et al. (4) and Samet and
organic and inorganic compounds, oxidants, change has been noted since 1990. From Cohen (5).
and acids, and contributes heavily to particulate 1988 to 1995 the average annual mean con- Disclaimer. The contents of this review are the
air pollution in most urban settings. The centration of PM1o fell by 17% (8). The responsibility of the author, and do not necessarily
reflect the views and policies of the Health Effects
radionudides are emitted as a result of the com- PM1o standard was augmented recently by a Institute or its sponsors.
bustion of fossil fuels as well as from mining PM2.5 standard, so there are few data on Received 23 September 1999; accepted 23
operations, and the asbestos fibers result from trends in this pollutant. December 1999.

Environmental Health Perspectives * Vol 108, Supplement 4 * August 2000 743

Table 1. Selected known carcinogens in urban and rural ambient air. On the basis of evidence from animal
Substance Urban air Rural air experiments and epidemiologic studies of
occupationally exposed groups, IARC consid-
Inorganic particulates (ng/m3) ers diesel exhaust to be a probable human car-
Arsenic 2-130 < 0.5-5
Asbestos 10-100 cinogen (IARC classification 2A) (22),
Chromium 5-120 < 1-10 although the mechanism by which exposure to
Nickel 10-1000 < 10 diesel exhaust might produce lung cancer in
Radionuclides (Ci/m3) humans remains to be determined. The evi-
210Pb 1 x 10-15-30 x 10~15 5.5x 10-15-10 x 1015 dence for the carcinogenicity of diesel exhaust
212Pb 0.1 x lO-15 - 4 x 10-15 0.03 x 10-15 -0.06 x 10-15 has recently been extensively reviewed (19,23).
222Rn 20 x 10-125- 1,000 x 10-12 0.1 x 10-12 20 x 10-12
Gaseous and particulate organic species (ng/m3) 1,3-Butadiene
Benzene 5-90 1,3-Butadiene is a volatile organic compound
Benzo[a]pyrene 1-50 employed since the 1930s in the production
Benzene-soluble organics 1,000-2,000 200-300
of synthetic rubber. It is also emitted in auto-
Adapted from IARC ( 10. motive exhaust and has been classified by
IARC as a probable human carcinogen (2A)
reactions in the outdoor environment physically diverse group of pollutants derived on the basis of results of animal inhalation
(11-13). Neither the contribution of these lat- from sources as diverse as crustal dust and sea experiments, which indicated increases in
ter, secondary compounds, to total outdoor spray and from the combustion of diesel fuel tumors at multiple sites, including the lung.
levels of POM nor their relative mutagenicity (18). As such, it is not possible to address Epidemiologic studies of occupationally
and carcinogenicity are well understood. generically the carcinogenicity of particles. exposed populations (rubber workers and
Although the combustion of fossil fuels is Largely because carbonaceous particles pro- butadiene monomer production workers)
a ubiquitous source of POM in the urban duced by the combustion of fossil fuels are in have consistently observed increases in
outdoor environment, it is not the only the respirable range (generally < 1.0 pm in hematopoietic cancers but not in cancers of
source of human exposure to POM, and for diameter) and known human carcinogens the respiratory system (20,24).
some individuals it may not be the predomi- such as PAHs are adsorbed to their surfaces,
nant source. Other human exposure to POM attention has focused on combustion-source Aldehydes
comes from inhaling wood and tobacco particles in urban air as potential lung car- Various aldehydes classified as hazardous air
smoke and from the diet (e.g., from the con- cinogens. However, recent evidence from pollutants by the U.S. EPA (e.g., formalde-
sumption of grilled meat). Unfortunately for animal experiments showed that rats exposed hyde, acetaldehyde) are present in urban ambi-
the conduct of epidemiologic research, there to high levels of relatively pure carbon parti- ent air largely due to the combustion of
are currently no validated markers of expo- cles developed lung tumors at the same rate gasoline and diesel fuel (24). Formaldehyde
sure to POM from specific sources, either in as rats exposed to diesel exhaust particles has been classified by IARC as a probable
outdoor air or biologic material. when each was compared to unexposed con- human carcinogen (2A) (25), on the basis of
Urban air contains a mixture of polycyclic trols, suggesting that particles per se might, evidence from animal experiments and epi-
organic compounds, but certain specific con- under some conditions, be carcinogenic. The demiologic studies in occupational groups of
stituents, such as benzo[a]pyrene, have been relevance of these findings for humans is exposure-related excess nasal and nasopharyn-
extensively studied and are known to be car- controversial (19). geal cancer. There is no consistent evidence
cinogenic. Benzo[a]pyrene has been used fre- The combustion of fossil fuels for power that occupational exposure to formaldehyde is
quently as a surrogate or marker for generation and transportation produces associated with increased lung cancer risk (25).
combustion source air pollution in epidemio- gaseous pollutants such as sulfur dioxide The combustion of alternative fuels, such as
logic studies and for risk assessment. The lit- (SO2) and oxides of nitrogen (NOx) that are methanol and oxygenated fuels containing the
erature on cancer risk in relation to converted into fine particulate air pollution in additive methyl tertiary butyl ether, results in
occupational and environmental exposure to the atmosphere. Epidemiologic studies pro- greater aldehyde emissions and contributes to
PAHs has recently been reviewed by Boffetta vide no consistent evidence of increased lung increased ambient concentrations in locales
et al. (14), who concluded that PAHs are cancer risk from occupational exposure to where they are widely used (24).
associated with increased lung cancer risk in a SO2; however, the International Agency for
variety of occupational settings and with Research on Cancer (IARC) has classified Epidemiologic Evidence
increased lung cancer risk in urban popula- strong sulfuric acid aerosol as a known human on Outdoor Air Pollution
tions. Mixtures of polycyclic compounds carcinogen based on epidemiologic findings of and Lung Cancer
encountered in occupational settings, such as increased lung and laryngeal cancer in heavily Several lines of epidemiologic research provide
coke-oven workers in the steel industry and exposed occupational groups (20). evidence about the association of ambient air
coal gasification workers (15,16), also are pollution with lung cancer occurrence in the
known to cause increased occurrence of lung Diesel general population: a) studies comparing lung
cancer in exposed workers (17). The levels of Diesel exhaust is a ubiquitous component of cancer risk in migrants to areas having differ-
POM encountered in the outdoor urban urban outdoor air pollution throughout the ing lung cancer risk from the native country;
environment, however, are substantially less world, although few studies have estimated its b) studies comparing cancer rates among
than those encountered in heavily exposed proportional contribution. In one of the few urban and rural populations; c) studies of pop-
occupational settings. studies that estimated the proportional contri- ulations residing near specific point sources of
bution of diesel exhaust to outdoor air pollu- air pollution; d) case-control and cohort
Partides tion, Cass and Gray (21) estimated that diesel studies of lung cancer occurrence in the gen-
Like POM, particulate air pollution is not a exhaust contributed 7% of the fine particulate eral urban population; and e) studies compar-
single entity but rather a chemically and (< 2 pm) in the Los Angeles air basin in 1982. ing the relative frequency of biologic markers

744 Environmental Health Perspectives * Vol 108, Supplement 4 * August 2000

of air pollution exposure and of air pollution Municipal solid waste incinerators emit heavy relative risks (compared to national incidence
exposure and genetic damage. metals (e.g., lead), cadmium, PAHs, organic rates) were adjusted for age, sex, geographic
compounds (such as dioxins), and acidic gases region, and an index of socioeconomic status.
Migrant Studies and Urban-Rural (34). Unfortunately, these sources of air pol- For several types of cancers (stomach, colorec-
Comparisons lution are often located in or near poor tal, liver, lung), excess relative risk was
Studies of migrants reviewed elsewhere (26) working-class communities whose residents inversely related to distance of the residence
provide limited evidence in support of the may, for a variety of reasons, be more suscep- from the incinerator. Lung cancer relative
hypothesis that air pollution is associated with tible to the effects of these pollutants (35). risks (95% confidence intervals [CIs]) were
lung cancer risk. Migrants from countries with In 1990, Pershagen (36) reviewed the 1.08 (1.07, 1.09) and 1.06 (1.05, 1.07) for
higher rates of lung cancer and higher levels of available epidemiologic studies of lung cancer residences 0-3 km and 0-7.5 km, respec-
air pollution to countries with lower air pollu- occurrence and residential proximity to tively, from the incinerator. However, Elliott
tion levels tend to develop lung cancer at rates industrial point sources of air pollution. and colleagues (38) also observed equal eleva-
higher than those of the new country of resi- Eleven studies estimated lung cancer risk tions in lung cancer risk in the areas proximal
dence, suggesting that prior exposure was a risk associated with proximity to nonferrous metal to the incinerators before construction of the
factor. However, incomplete control for the smelters. Of these, flve ecologic studies facilities, leading them to conclude that resid-
effects of smoking and occupational exposure observed relative risks in males of between 1.2 ual confounding by unmeasured characteris-
may account for these findings. and 2.0, but only one study accounted for tics of the postal codes accounted for the
The earliest studies of air pollution and employment at the smelter itself, and data on apparent associations with proximity to the
lung cancer contrasted lung cancer rates of smoking were not available. These studies did incinerators.
urban and rural populations. Most studies not consistently observe elevations in risk Ecologic studies, such as those of Archer
found overall excesses on the order of among women. Six case-control studies pre- (37) and Elliott et al. (38), have generally
30-40% in the urban areas and larger relative sented conflicting results; several showed no observed relative excesses of lung cancer in the
excesses among nonsmokers. The attribution association with residential proximity and did more polluted areas of similar or slightly higher
of these results to differences in air quality was not account for either employment at the magnitude than the urban-rural studies (5).
strengthened by evidence of urban-rural dif- facility or smoking habits. Two studies that However, because incidence, exposure, and
ferences in ambient levels of carcinogens such did account for these factors observed relative covariate data are all measured on the aggre-
as benzo[a]pyrene and by the frequent persis- risks in males of 1.6 and 2.0. Ecologic studies gate, or ecologic, level, it is difficult to account
tence of the urban-rural differences after of residential proximity to diverse industrial adequately for intraindividual and between-
adjustment for cigarette smoking. Doll and sources (e.g., petrochemical plants and steel area differences in other risk factors (39).
Peto (27), in their widely cited monograph, mills) generally observed increased rates of
The Causes of Cancer, cast doubt on the causal lung cancer but were unable to control for Case-Control and Cohort Studies
role of air pollution because early research had confounders at the individual level, i.e., such of Lung Cancer Occurrence
not accounted for the effects of urban dwellers factors as cigarette smoking and employment in the General Urban Population
who started smoking at younger ages as ciga- at the industrial facility itself Cohort and case-control studies have the
rette smoking became increasingly prevalent Taking advantage of a "natural experi- advantage of offering information on potential
in the early 20th century. However, Dean ment" in which air pollution levels changed confounding and modifying factors such as
(28) controlled for the age at which smoking in a relatively rapid and clear-cut fashion over cigarette smoking. Case-control studies pro-
began and found that the urban-rural gradi- time, Archer (37) analyzed respiratory cancer vide an efficient approach to estimating the
ent persisted. Recent lung cancer incidence mortality in two counties in Utah with very relative risk of lung cancer in relation to air
and mortality data continue to show evidence low smoking rates. The counties were similar pollution exposure without having to collect
of urban-rural differences (29,30) and other in many respects, with low and nearly equal information on an entire cohort or study pop-
studies document contemporary urban-rural respiratory cancer mortality rates, until a steel ulation. In the case-control design, cases of
gradients in the mutagenicity of airborne par- mill constructed during World War II caused lung cancer that have occurred in the popu-
ticulate matter across the United States substantial increases in air pollution in one lation are ascertained and classified according
(31,32). Nonetheless, the urban factor may county. The subsequent differences in inci- to exposure to outdoor or indoor air pollu-
reflect influences instead of, or in addition to, dences of lung cancer cases were substantial tion; a sample of the study population-the
outdoor air pollution; these could include within about 15 years after the increase in air controls is selected and similarly classified
indoor air pollution, patterns of migration, pollution and have persisted. Presumably the according to exposure. An estimate of the rela-
occupational exposures, or factors related to prevalence of smoking remained constant, or tive risk can then be calculated from these
population density. at least equal, in the two counties. A third data. However, the validity of that estimate
neighboring county, unaffected by pollution depends on both the cases and the controls
Studies of Populations Residing Near from the steel mill but with a population with being selected independent of their exposure.
Point Sources of Air Pollution higher smoking rates had higher lung cancer Cohort and case-control studies of lung
Residential proximity to industrial point rates than either of the other two counties, cancer are usually retrospective, i.e., lung can-
sources of air pollution is a potential source of underscoring the profound effects of cigarette cers have already occurred when the study is
exposure to known or suspected carcinogens, smoking on lung cancer risk. conducted. Therefore, a strategy is needed for
as noted previously. Fossil fuel-fired (i.e., Elliott and colleagues (38) recently estimating air pollution exposure using his-
coal, oil, natural gas) electrical power plants reported the results of an ecologic study of torical information. Investigators have gener-
emit known or suspected carcinogens (12), cancer incidence among 14 million people ally linked available air pollution monitoring
including metals such as chromium and living near 72 municipal solid waste incinera- data with residential histories to provide esti-
nickel, radionuclides such as radon and ura- tors in Great Britain. Cancer rates and resi- mates of long-term exposure to air pollution.
nium, and POM such as benzo[a]pyrene. dential proximity to the incinerators were The exposure of an individual to carcinogens
Nonferrous metal smelters emit inorganic measured at the postal code level (which is may occur in multiple microenvironments,
arsenic and other metals and SO2 (33). roughly analogous to neighborhood); the and therefore may be difficult to characterize

Environmental Health Perspectives * Vol 108, Supplement 4 * August 2000 745

for the purpose of epidemiologic analysis. For excess lung cancer risk for a difference in fine was associated with air pollution when sulfate
example, early lung cancer studies often particulate mass equal to that of the most pol- particulate was used as the index of air pollu-
defined exposure to outdoor air pollution in luted versus the least polluted city. However, tion exposure but not when fine particle mass
terms of urban and rural location of resi- the exposure of cohort members to fine par- (available for only 50 cities and approximately
dence. More recent investigations have used ticulate air pollution (estimated as the average half the study population) was used, as in the
crude indicators of cumulative exposure to air level of fine particulate mass over the entire Six-Cities study (45). This discrepancy did not
pollution such as duration of residence in an risk period) was assumed to have been con- appear to be due to differences in air monitor-
area characterized by a particular level of pol- stant throughout their lives, and exposures to ing data across the 151 metropolitan areas and
lution. The level is usually derived from rou- other risk factors (such as cigarette smoking) remains unexplained.
tinely collected air monitoring data from one, were assumed to have been stable over the The Adventist Health Study on Smog
or at most several, stationary monitoring sites. approximately 16-year period of follow-up. reported by Beeson and colleagues (47) fol-
Although these approaches may identify most In addition to the Six-Cities study (45) lowed a cohort of Seventh Day Adventists, res-
truly exposed subjects, they tend to classify two other U.S. prospective cohort studies ident in southern California, whose extremely
some truly unexposed subjects as exposed. observed increased relative risks of lung cancer low prevalence of smoking and uniform (and
An individual's exposure may also change associated with exposure to air pollution. Pope relatively healthy) dietary patterns reduce the
over time due to, for example, trends in air and colleagues (46) linked ambient air pollu- potential for confounding by these factors.
pollution levels or subject mobility. Although tion data from 151 U.S. metropolitan areas Lifetime exposure to a range of air pollutants
lung cancer is a disease of late adulthood, if with risk factor data for 552,138 adults was estimated for each study subject, using
air pollution plays a causal role, it is possible enrolled in the American Cancer Society ambient air quality measurements collected by
that exposures to air pollution in earlier peri- Cancer Prevention Study II (CPS-II), and the U.S. EPA and the state of California, resi-
ods may be important. Few, if any, current monitored for vital status from 1982 to 1989. dence histories, and questionnaires. Both
studies of air pollution and lung cancer have Using multivariable regression, the investiga- PMto and ozone were associated with lung
assessed the role of time-varying air pollution tors controlled for individual differences in cancer (Table 2), but the 5-fold increase in
exposure on lung cancer occurrence. age, sex, race, cigarette smoking, pipe and cigar lung cancer rates associated with cumulative
Misdassification of exposure can spuriously smoking, exposure to passive cigarette smoke, exposure to PM1o is not consistent with the
elevate or diminish estimates of effects. When occupational exposure, education, body mass body of current research and is difficult to
misdassification rates are the same for exposed index, and alcohol use. Lung cancer mortality explain even in a population with relatively few
cases and controls and for unexposed cases and
controls (i.e., nondifferential misdassification),
estimates of effect are in most cases attenuated Table 2. Epidemiologic studies of outdoor air pollution and lung cancer.
(40Q. When the risk of lung cancer increases Rate ratio
directly and monotonically with exposure Study Locale Exposure classification (95% Cl)
(sometimes referred to as a dose response), Ecologic
nondifferential misclassification of exposure Henderson (77) Los Angeles, CA Polycyclic aromatic hydrocarbon levels 1.3
can obscure this pattern (41,42. Most studies by geographic area
attempt to collect data on other lung cancer (TSP weekly mean 96-116 pg/m3)
risk factors such as cigarette smoking, which Buffler (78) louston, TX TSP levels by census tract 1.9
could confound the air pollution relative risks. (RR evaluated at 16.12 pg/M3)
Errors in the measurement of potential con- Archer (37) Utah Mean levels of TSP by county 1.6
founders can have more serious consequences, (Mean TSP in high air pollution area 85 pg/M3)
producing bias in either a positive or negative Case-control
direction even if air pollution exposures are Pike (81) Los Angeles Residence in high pollution (benzo[alpyrene) area 1.3a
estimated with relatively little error (43). (TSP weekly mean 96-116 pg/M3) (NA)
Misclassification of exposure and potential Vena (82) Buffalo, NY > 50 years residence in high TSP areas (80-200 pg/M3) 1.7a
effect modifiers can also create bias and impre- (1.0-2.9)
cision in estimates of interaction (44). Jedrychowski (83) Cracow, Poland Residence in high TSP (> 150 pg/m3) and S02 1.5a
Most published cohort and case-control (> 104 pg/m3) areas (1.1-2.0)
studies found relative increases of lung cancer Katsouyanni (84) Athens, Greece Lifelong residence in high pollution areas (soot 1.1a
risks after adjustment for age, smoking, and levels as high as 400 pg/m3) (NA)
occupational exposure similar to those Barbone (48) Trieste, Italy Residence in areas with high levels of particle 1.4a
observed in the urban-rural and ecologic deposition (> 0.298 g/m2/day) (1.1-1.8)
studies (Table 2). Dockery and colleagues Nyberg et al. (49 Stockholm, Long-term (2 30 years) residential exposure to 1.3a
(45) reported the results of a cohort study of Sweden traffic-related NO2 (0.9-1.9)
8,111 adults living in six U.S. cities. Cohort Cohort
members were followed for between 14 and Beeson (47 California RR evaluated at mean PM10 IOR 24 pg/m3 (males) 5.2a
16 years and their mortality was ascertained (1.9-13.9)
RR evaluated at 8-hr mean 03 IOR 100 ppb (males) 1.7a
through 1989. Lung cancer relative risks were (0.7-3.8)
estimated with respect to average levels in Dockery (45) 6 U.S. cities Residence in high fine particulate pollution areas 1.4a
each city for various components of air pollu- (RR evaluated at PM2.5 = 18.6 pg/M3) (0.8-2.3)
tion, including total and fine particulate Pope (48) 151 U.S. cities Residence in high sulfate particulate pollution areas 1.4a
mass, ozone, and sulfate particles. After (RR evaluated at sulfate = 19.9 pg/M3) (1.1-1.7)
adjustment for differences in age, sex, ciga- Residence in high fine particulate pollution areas 1.0a
rette smoking, obesity, and education among )RR evaluated at PM2.5 = 24.4 pg/M3) (0.8-1.3)
cohort members, researchers observed a 37% IOR, interquartile range. &Denotes studies that controlled for cigarette smoking. Table modified from Samet and Cohen (5).

746 Environmental Health Perspectives * Vol 108, Supplement 4 * August 2000

other risk factors. Unfortunately, the authors lung cancer risk associated with air pollution. adducts with PAHs were associated with
do not present the actual incidence rates that Potential biomarkers for lung cancer risk workplace exposure (54,55).
might have permitted this issue to be evalu- include actual levels of the putative carcino- Studies have also been conducted of
ated directly. gen in biologic materials, DNA adducts of biomarkers in traffic police workers and bus
Barbone et al. (48) conducted a case- potential carcinogens or metabolites, and drivers. In a study of traffic police workers in
control study of lung cancer mortality from antibodies against such adducts (50). Italy, levels of micronudei in peripheral blood
1979 to 1986 and exposure to ambient air Biomarkers of exposure to respiratory car- lymphocytes were not increased in the police
pollution in Trieste, Italy. Exposures for 755 cinogens provide an intermediate outcome workers in comparison with controls (56).
individuals who died of lung cancer and 755 for investigation that may prove valid surro- Urinary excretion of 1-hydroxypyrene, how-
controls who died from other causes were gates for risk. Levels of adducts, for example, ever, did appear to be a useful biomarker for
estimated as the average level of particle depo- may prove to be predictors of risk and may exposure to air pollution by PAHs. In contrast,
sition (g/m2/day) for their neighborhood of bridge from animal models and in vitro assays bus drivers working in central Copenhagen,
residence between 1972 and 1979. After con- to human risk. These methods may enable Denmark, had higher levels of PAH-DNA
trolling for the effects of age, smoking, and epidemiologists to reduce bias from misclassi- adducts in comparison with controls (57).
occupation, a 40% increase in the lung cancer fication of exposure, but these methods are in
mortality rate (RR = 1.4; 95% CI 1.1, 1.8) their infancy and will require careful study Air Pollution and Lung Cancer
was observed among those who resided in and more extensive application before their in Less-Developed Countries
neighborhoods with the highest levels of true utility is known. Mounting evidence indicates that popula-
particle deposition (> 0.298 g/m2/day). Studies of biomarkers of air pollution tions in less-developed countries may have
Case-control studies based only on deceased exposure and intermediate effects related to exposure to indoor and outdoor environ-
individuals present certain well-known inter- carcinogenicity have focused largely on a single ments that rival or even dramatically exceed
pretative problems. Here, 55% of the con- group of carcinogens present in urban air: the those found in developed Western countries.
trols had died from cardiovascular disease, PAHs, of which benzo[a]pyrene is the proto- Indoor air pollution from coal combustion
which itself has been associated with particu- type. The concentration of benzo[a]pyrene can and cooking fumes has been linked to
late air pollution (45). If the association with be measured in the air and a number of studies increased lung cancer risk in homes in
cardiovascular mortality is valid, and the only have examined the association between expo- China and Hong Kong (58). Rising pollu-
feature that distinguished residential areas was sure to benzo[a]pyrene and other PAHs and tion of outdoor air in the mega-cities of the
the level of air pollution, then the observed marker levels, including PAH-DNA adducts, developing world may also pose a risk for
relative risk would be an underestimate of the sister chromatid exchange, chromosome aber- lung cancer.
true relative risk. If, however, other features rations, and oncogene proteins; associations Within the less-developed countries, the
of residential area (e.g., social class) are also generally have been positive. For the most part, highest exposures, particularly among
associated with cardiovascular mortality, then exposure to PAHs has been studied in occupa- women, have been to indoor air pollution
the air pollution relative risk in this study tionally exposed workers, including foundry from the combustion of coal and biomass
may be an overestimate. workers and traffic police, but it also has been fuels for cooking and heating (59). For exam-
Nyberg and colleagues (49) recently studied in residents of highly polluted urban ple, typical concentrations of coal smoke in
completed a case-control study of air pollu- environments, largely in Eastern Europe. rural homes in China exceeded 500 pg/m3
tion and lung cancer in the general popula- Perera and colleagues (51,52) investi- and frequently exceeded 1 mg/m3. Smith and
tion of Stockholm, Sweden (the LUCAS gated biomarkers in residents of a highly Liu (58) recently reviewed the epidemiologic
study) that attempted to identify the sources industrialized region of Silesia in Poland. literature on indoor air pollution and lung
of urban air pollution most strongly associ- Levels of benzo[a]pyrene were markedly ele- cancer in the developing countries and found
ated with lung cancer. The study included all vated due to industrial activity and coal com- consistent evidence of increased rates of lung
lung cancers (n = 1,042) that occurred among bustion for residential heating. Airborne cancer associated with indoor cooking and
male residents of Stockholm, 40-75 years of particulate matter collected in this region was heating with coal in studies done largely in
age, from 1985 to 1990. Historical measure- subjected to assays for mutagenicity and China. A much smaller group of studies
ments of SO2 and NO2/NO. were combined found to have genotoxic activity in a variety revealed no consistent association of lung
with residence histories to provide estimates of short-term tests. Compared with controls cancer with indoor use of biomass fuels.
of long-term exposure to air pollution from from a less polluted region, persons residing Mounting levels of urban air pollution,
fuel combustion from residential heating and in Silesia had significant increases in from local stationary and, increasingly,
traffic, respectively. Exposure for 30 years to carcinogen-DNA adducts, sister chromatid mobile sources (60) are recognized as an
NO2 in the highest decile was associated with exchange, and chromosomal aberrations. important environmental problem by inter-
a relative risk of 1.3 (95% CI 0.9, 1.9) after They also showed a doubling of the fre- national public health and economic agen-
adjustment for cigarette smoking, occupa- quency of ras oncogene overexpression. cies. In the cities of the poorest countries,
tional exposure, and radon exposure, and the These results indicate the potential to esti- the Global Environmental Monitoring
relative risk in this decile after the 21- to 30- mate lung cancer risk by combining informa- System of the World Health Organization
year induction time was 1.6 (95% CI 1.1, tion on concentrations of carcinogens in observed average ambient concentrations of
2.4). No positive associations were reported outdoor air with levels of adducts in exposed total suspended particles of 300 mg/m3 (61),
for SO2. persons; the levels of biomarkers could then although levels in locales where coal is used
be used to predict cancer risks once the bio- for fuel, such as poor communities in South
Studies of the Relative Frequency markers are validated (53). Workers in cer- Africa, may exceed 1 g/m3. Although there
of Biologic Air Pollution Exposure tain occupations, including foundries or jobs are currently few relevant investigations, a
and Genetic Damage involving heavy vehicle exhaust exposure, case-control study in Shenyang, China,
Exposure biomarkers-that is, indicators of have also been investigated for levels of bio- observed a 2-fold increase in lung cancer risk
exposure or dose measured in biologic mater- markers of exposure to PAHs. In a study of after adjustment for age, education, and
ial-offer a new approach to quantifying the foundry workers in Finland, levels of DNA smoking among residents in smoky areas of

Environmental Health Perspectives * Vol 108, Supplement 4 * August 2000 747

the city and a 50% increase among those in from the combustion of fossil fuels, observed air pollutant if only because, in contrast to Rn
somewhat or slightly smoky areas (62). in studies of varied design and in diverse set- studies in which exposure assessment methods
tings, suggest that such exposures may cause to a single pollutant are relatively uniform,
Estimating the Impact of Air Pollution small relative increases in lung cancer rates. investigators used a wide variety of indices to
on Lung Cancer Rates This interpretation is consistent with studies characterize the exposure of study subjects to
Estimation of the magnitude of the contribu- of other types of exposure to combustion- the complex mixture of ambient air pollution.
tion of air pollution to lung cancer occur- source pollution such as occupational expo- Without improved epidemiologic methods,
rence at contemporary levels of air pollution sures and exposures to environmental tobacco however, even large studies may fail to inform.
poses a major challenge. Samet (63) recently smoke. Errors in the measurement of air pol- Current development of biologic markers of
reviewed the issue of risk assessment of air lution exposure and in the measurement of exposure to and molecular effects of PAHs rep-
pollution exposure. Historically, estimates of other risk factors including cigarette smoking resents one approach to improving epi-
the population attributable risk of lung continue to limit our ability to quantify the demiologic methods. Markers of genetic
cancer due to outdoor air pollution have used magnitude of the excess lung cancer risks susceptibility are also needed. In addition and
diverse approaches and produced variable associated with air pollution. of equal importance, methods for the retro-
estimates. Basing their estimate on past and Relative to cigarette smoking, the excess spective estimation of long-term exposure to
then current estimates of benzo[a]pyrene in lung cancer risk associated with ambient air air pollutants should be developed and tested
urban air and extrapolation from occupa- pollution is small. Nonetheless, given the so that large case-control and retrospective
tional studies of PAH-exposed workers, Doll ubiquity of combustion-source ambient air cohort studies can be feasibly conducted.
and Peto (22) estimated that less than 1% of pollution exposure, the contribution of this Recent approaches to developing such esti-
future lung cancer would be due to air pollu- exposure across a population may be of public mates have combined time-activity informa-
tion from the burning of fossil fuels. They health importance even if exposure to ambi- tion, including long-term residential histories,
did note, however, that perhaps 10% of then- ent air pollution causes but a small propor- with data from national aerometric databases
current lung cancer in large cities might have tion of annual lung cancer mortality (which such as the Aeromatic Information Retrieval
been due to air pollution. In 1990, the U.S. currently totals approximately 150,000 System (AIRS) database maintained by the
EPA (64) estimated that 0.2% of all cancer, deaths annually in the United States). U.S. EPA (49,68,6p). Airport visibility data,
and probably less then 1% of lung cancer, Direct epidemiologic observation of routinely collected for several decades across
could be attributed to air pollution. This esti- exposed populations may yet provide better the United States, may also provide a historical
mate was obtained by applying the unit risks information for evaluating the magnitude of record of aerometric data that could be used to
for more than 20 known or suspected human outdoor air pollution-related excess lung estimate past levels of outdoor particulate air
carcinogens found in outdoor air to estimates cancer, but because the expected relative pollution (84). This effort should include
of the ambient concentrations and numbers effect of air pollution is likely to be weak in development of methods to characterize, quan-
of persons potentially exposed. The unit risks many settings, new studies that could better tify, and adjust for exposure measurement
were derived either from animal experiments guide policies for protection of public health error. For lung cancer, urban and relatively
or extrapolation from studies of workers will face considerable challenges. unpolluted areas with established population-
exposed to higher concentrations. One group In general, large-scale epidemiologic based tumor registries might be targeted.
based their estimates on direct observation of studies of air pollution and lung cancer will Although most case-control and cohort
populations exposed to ambient levels of air be needed if we are to obtain sufficiently studies have tried to address confounding due
pollution. Karch and Schneiderman (65), informative data, but it is not obvious that to cigarette smoking and occupation, virtually
using data from the American Cancer Society such studies are feasible. Assessing their feasi- none have addressed possible bias due to the
(CPS-I) study and U.S. Census data, esti- bility is a key research need. measurement errors in exposure and covari-
mated that the urban factor accounted for Large numbers of cases will be necessary to ates. Such bias, even if it is nondifferential,
12% of lung cancer in 1980. They predicted measure accurately and precisely the small rel- can produce either spuriously high or low esti-
that 1980 levels of TSP (approximately 60 ative excesses that have been observed and to mates of the lung cancer rate ratio in multi-
pg/m3) would be associated with a lung measure the joint effects of air pollution and variable data (40). The problem is that few if
cancer rate ratio of 1.32, slightly less than the other factors such as occupation and smoking. any studies have collected the data necessary
47% increase observed at an approximate The example of residential radon (Rn) pro- to quantify this bias, or often even to deter-
55.8 pg/m3 level of TSP in the Six Cities vides a useful analogy. Lubin et al. (67) con- mine its likely direction. These problems are
study (66). sidered the feasibility of conducting valid and particularly serious when the relative effects
Each attributable risk estimate cited above precise case-control studies of lung cancer and are small. Future studies should develop
is subject to considerable uncertainty because residential exposure to Rn and its decay prod- methods and collect data that can be used to
of a lack of knowledge about both the relative ucts given the expectation of small relative quantify exposure measurement error and
magnitude of the effect and the proportion of effects, errors in the measurement of exposure, compute adjusted effect estimates. Some spe-
the population exposed, but there seems to be and other sources of bias such as subject cific areas for future research should include:
no compelling argument to prefer estimates mobility. All these conditions apply in the Identification of the pollutants and
based on extrapolation from animal experi- case of ambient air pollution, except that the pollution sources associated with increased
ments or occupational studies to direct epi- problems posed by measurement error in the occurrence of lung cancer. Studies in single
demiologic observation of the general air pollution case are, if anything, more seri- locales where levels of individual pollutants are
populations at risk if valid and reasonably ous. Lubin and colleagues concluded that the highly correlated will generally be unable to
precise epidemiologic results are available. required study size might well be infeasible estimate the contributions of specific pollu-
and recommended pooling the data from tants or sources, although arguably this may
Conclusions and existing studies with the largest number of be possible in certain situations such as the
Research Needs cases and most precise exposure estimates. It is Swedish case-control study discussed previ-
The repeated associations between lung not at all clear, however, that this solution ously (49). New designs and statistical meth-
cancer occurrence and air pollution, chiefly could be applied to studies of a single ambient ods for air pollution studies may provide

748 Environmental Health Perspectives * Vol 108, Supplement 4 * August 2000

additional insights. Navidi et al. (70) and chronic and/or episodic inflammatory insults air pollution that may contribute to increased
Prentice and Sheppard (71) have described such as those that characterize asthma have occurrence of lung cancer.
hybrid studies that combine ecologic-level been hypothesized to play a role in lung As a greater proportion of the world's
contrasts of air pollution effects between cities carcinogenesis. population moves from rural communities'to
with individual-level data on covariates, com- Ideally, longitudinal observation of large the rapidly expanding and highly polluted
bining the strengths of both ecologic and populations over decades would be required to cities of Asia and the Southern Hemisphere,
individual-level studies. Studies using these determine the possible role of long-term expo- there is a need to address the large gap in epi-
designs could contrast the effect on lung sure to air pollution in the pathogenesis of demiologic research on air pollution and lung
cancer of exposure to the pollutant mixtures chronic respiratory disease and the subsequent cancer in the developing world.
of different cities while effectively controlling development of lung cancer. If the current These studies will present even greater
confounding by cigarette smoking, diet, or high level of interest in the effects of long-term challenges than those in the industrialized
other factors, and adjusting for exposure mea- exposure to particulate air pollution on chronic West. In addition to the generic problem of
surement error. Such a study might be carried disease incidence and mortality results in either estimating long-term exposure to air pollu-
out in the United States using the combined new cohort studies or the retro-fitting of exist- tion, the ambient air pollution mixture in
resources of the Surveillance, Epidemiology ing cohorts with air pollution data, such longi- urban centers in the developing countries is
and End Results (SEER) cancer registry or tudinal observations may be possible. changing, due in part to the increase in auto-
other population-based cancer registries, and Measurement ofthe interaction ofambient mobile traffic. Characterizing these changes as
the AIRS and other air pollution databases air pollution with other known or suspected they occur over time, including choosing and
maintained by the U.S. EPA. If the air pollu- causes of lung cancer, ie., cigarette smoking, measuring indicator pollutants for different
tion mixtures in various U.S. population cen- occupation, and diet. The effect of air pollu- pollution sources, requires careful planning.
ters could be characterized both in terms of tion on lung cancer occurrence may depend, In addition, the current increases in cigarette
physical and chemical constituents and perhaps critically, on other factors, such as cig- smoking in the developing world (61), and
sources of major constituents, then it might be arette smoking, genetic predisposition, diet, their thoroughly predictable consequences,
possible to retrospectively estimate long-term occupational exposures, and social class, to will complicate interpretation of future
pollutant- and source-specific exposures of cause lung cancer. Some reviewers have noted studies of air pollution and lung cancer.
study subjects. a greater-than-additive relationship between air
Studies of occupational cohorts have pollution and cigarette smoking that would REFERENCES AND NOTES
historically provided important information imply synergism (26,65). Estimates of the
1. Doll R, Hill AB. A study of the aetiology of carcinoma of the
about the carcinogenicity of specific pollu- magnitude of the effect of joint exposure to lung. Br Med J 2:740-748 (1950).
tants such as diesel exhaust, to which the gen- ambient air pollution and cigarette smoking 2. White C. Research on smoking and lung cancer: a landmark in
eral population is exposed, as discussed have been reported or can be derived from sev- the history of chronic disease epidemiology. Yale J Biol Med
eral studies (65), and although these results 63:29-46 (1990).
previously. Although epidemiologists have, 3. Tomatis L, ed. Air Pollution and Human Cancer. In: European
for reasons of study power, traditionally appear to suggest a greater-than-additive rela- School of Oncology Monographs. Berlin:Springer-Verlag, 1990.
sought out occupational groups exposed to tion between cigarette smoking and ambient 4. Cohen AJ, Pope CA Ill, Speizer FE. Ambient air pollution as a
levels of pollution well in excess of ambient air pollution (anywhere from 20 to 45% of risk factor for lung cancer. Salud Publica Mex 39(3):349-355
(1997).
levels, there may be a benefit to focusing cases attributable to joint exposure), they are 5. Samet JM, Cohen AJ. Air pollution and lung cancer. In: Air
future research on occupational groups who subject to error from inaccurate measurement Pollution and Health (Holgate S, Koren H, Maynard R, Samet
are exposed to levels of pollution more com- of both air pollution exposure and cigarette JM, eds). London:Academic Press, 1999;841-864.
6. Council for Environmental Quality. Eleventh Annual Report of
parable (in terms of intensity) to those of the smoking, and to substantial imprecision due to the Council on Environmental Quality. Washington, DC:U.S.
general population. With regard to diesel small numbers of lung cancer cases among Government Printing Office, 1980.
exhaust, for example, studies of truck or taxi nonsmokers. 7. Daisey JM, Lioy PJ, Daisy JM, eds. Toxic Air Pollution-A
Comprehensive Study of the Non-Criteria Air Pollutants.
drivers, may combine the methodologic Although the combined effects of ambient Chemical Composition of Inhalable Particulate Matter:
strengths of studies based in an occupational air pollution and smoking have not been well Seasonal and Inter-Site Comparisons. Chelsea, Ml:Lewis
population with the generalizability of near characterized, the combined effects of smok- Publishers,1 988.
8. U.S. EPA. National Air Quality and Emissions Trends Report,
ambient-level exposures. ing with air pollution merits particular con- 1995. EPA 454/R-96-005. Research Triangle Park, NC:U.S.
Investigation of the etiologic mechanisms sideration because of the strength of smoking Environmental Protection Agency, 1996.
that might underlie air pollution's role in as a cause of lung cancer and the continued 9. Cass GR, Conklin MH, Shah JJ, Huntzicker JJ, Macias ES.
Elemental carbon concentrations: estimation of an historical
lung cancer occurrence, including its role in high, and in some cases growing, prevalence data base. Atmos Environ 18:153-162 (1984).
the association oflung cancer with preexisting of smoking in many parts of the world. 10. IARC. IARC Monographs on the Evaluation of Carcinogenic
respiratory disease. The risk of lung cancer is However, as illustrated by the example of res- Risks to Humans. Supplement 7: Overall Evaluations of Carcino-
increased among people with preexisting lung idential Rn noted previously, prohibitively genicity, an Updating of IARC Monographs. Lyon:lnternatonal
Agency for Research on Cancer, 1987.
disease (e.g., chronic obstructive pulmonary large studies, and unachievably accurate mea- 11. Greenberg A. Analyses of polycyclic aromatic hydrocarbons. In:
disease and asthma) and among those who surement of both air pollution and cigarette Toxic Air Pollution: A Comprehensive Study of Non-Criteria Air
have exhibited accelerated rates of decline in smoking may be required to provide valid Pollutants (Daisey JM, Lioy MJY, eds). Chelsea, Ml:Lewis
Publishers, 1987:93-118.
pulmonary function (72). Long-term exposure estimates of combined effects (76). 12. Natusch DF. Potentially carcinogenic species emitted to the
to air pollution has also been associated with Studies of the contribution of ambient air atmosphere by fossil-fueled power plants. Environ Health
low levels of lung function and chronic respi- pollution to lung cancer occurrence in less- Perspect 22:79-90 (1978).
13. Winer AM, Busby WF Jr. Atmospheric transport and transfor-
ratory symptoms in several cross-sectional and developed countries currently undergoing mation of diesel emissions. In: Health Effects of Diesel Engine
longitudinal studies of children (73-77). rapid urbanization. Current knowledge about Emissions: Characterization and Critical Analysis. Cambridge,
These reported associations suggest mecha- ambient air pollution and lung cancer is based MA:The Health Effects Institute, 1995.
14. Boffetta P, Jourenkova N, Gustavsson P. Cancer risk from occu-
nisms other than, or in addition to, the direct largely on the experience of populations of pational and environmental exposure to polycyclic aromatic
initiation or promotion of lung tumors, by Western industrialized nations. However, hydrocarbons. Cancer Cause Control 8:444-472 (1997).
which prolonged exposure to air pollution industrial and infrastructure development in 15. Doll R, Vessey MP, Beasley RWR, Buckley AR, Fear EC, Fisher
poorer countries has led to increases in urban REW, Gammon EJ, Gunn W, Hughes GO, Lee K, en al. Mortality
could increase lung cancer risk. For example,

Environmental Health Perspectives * Vol 108, Supplement 4 * August 2000 749

of gas workers. Final report of a prospective study. Br J Ind 39. Greenland S, Robins J. Invited commentary: ecologic studies- 63. Samet JM. Risk assessment and air pollution. In: Air Pollution
Med 29:394-406 (1972). biases, misconceptions, and counterexamples. Am J Epidemiol and Health (Holgate S, Koren H, Maynard R, Samet JM, eds).
16. Redmond CK. Cancer mortality among coke oven workers. 139)8):747-760 (1994). London:Academic Press, 1999;881-898.
Environ Health Perspect 52:67-73 (1983). 40. Hatch M, Thomas D. Measurement issues in environmental epi- 64. U.S. EPA. Cancer Risk from Outdoor Exposure to Air Toxics.
17. IARC. Industrial exposures in aluminum production, coal gasifi- demiology. Environ Health Perspect 101(suppl 4):49-57 (1993). Vol 1. Final Report. EPA-450/1-90-004a. Research Triangle Park,
cation, coke production, and iron and steel founding. In: IARC 41. Birkett NJ. Effect of nondifferential misclassification on esti- NC:U.S. Environmental Protection Agency, 1990.
Monographs on the Evaluation of Carcinogenic Risks to mates of odds ratios with multiple levels of exposure. Am J 65. Karch NJ, Schneiderman MA. Explaining the Urban Factor in
Humans: Polynuclear Aromatic Compounds. Vol 34, Pt 3. Epidemiol 136:356-362 (1992). Lung Cancer Mortality. A Report of the Natural Resources
Lyon:lnternational Agency for Research on Cancer, 1984. 42. Dosemici M, Wacholder S, Lubin JH. Does nondifferential mis- Defense Council. Washington, DC:Clement Associates, Inc.,
18. Koutrakis P, Sioutas C. Physico-chemical properties and mea- classification of exposure always bias a true effect toward the (1981).
surement of ambient particles. In: Particles in Our Air: null value? Am J Epidemiol 132:746-768 (1990). 66. Dockery D. Personal communication.
Concentrations in Health Effects (Spengler JD, Wilson R, eds). 43. Greenland S. The effect of misclassification in the presence of 67. Lubin JH, Samet JM, Weinberg C. Design issues in epidemio-
Cambridge, MA:Harvard University Press, 1996;1 5-40. covariates. Am J Epidemiol 112:564-569 (1980). logic studies of indoor exposure to Rn and risk of lung cancer.
19. Cohen AJ, Nikula K. The health effects of diesel exhaust: labo- 44. Garcia-Closas M, Rothman N, Lubin J. Misclassification in Health Physics 59(6):807-817 (1990).
ratory and epidemiologic studies. In: Samet JM, Cohen AJ. Air case-control studies of gene-environment interactions: assess- 68. Kunzli N, Lurmann F, Segal M, Ngo L, Balmes J, Tager IB.
Pollution and Lung Cancer. In: Air Pollution and Health, Vol 32 ment of bias and sample size. Cancer Epidemiol Biomarkers Association between lifetime ambient ozone exposure and pul-
(Holgate S, Koren H, Maynard R, Samet JM, eds). London: Prev 8:1043-1050 (1999). monary function in college freshmen-results of a pilot study.
Academic Press, 1999;707-748. 45. Dockery DW, Pope CA Ill, Xu X, Spengler JD, Ware JH, Fay ME, Environ Res 72:8-23 (1997).
20. IARC. Occupational exposures to mists and vapours from strong Ferris BG Jr, Speizer FE. An association between air pollution 69. Aerometric Information Retrieval Service Database.
inorganic acids; and other industrial chemicals. In: IARC and mortality in six U.S. cities. N EngI J Med 329(24): Washington, DC:U.S. Environmental Protection Agency.
Monographs on the Evaluation of Carcinogenic Risks to 1753-1759 (1993). Available: http.//www.epa.gov/airs/airs.html[cited 1999].
Humans, Vol 54. Lyon:lnternational Agency for Research on 46. Pope CA Ill, Thun MJ, Namboodiri MM, Dockery DW, Evans JS, 70. Navidi W, Thomas D, Stram D, Peters J. Design and Analysis of
Cancer, 1992. Speizer FE, Heath CW Jr. Particulate air pollution as a predictor Multigroup Analytic Studies with Applications to the Study of
21. Cass GR, Gray HA. Regional emissions and atmospheric con- of mortality in a prospective study of U.S. adults. Am J Respir Air Pollution. Technical Report No 49. Los Angeles,
centrations of diesel engine particulate matter: Los Angeles as Crit Care Med 151(3):669-674 (1995). CA:University of Southern California,1 993.
a case study. In: Health Effects of Diesel Engine Emissions: 47. Beeson WL, Abbey DE, Knutsen SF. Long-term concentrations 71. Prentice RL, Sheppard L. Aggregate data studies of disease risk
Characterization and Critical Analysis. Cambridge, MA:The of ambient air pollutants and incident lung cancer in California factors. Biometrika 82:113-125 (1995).
Health Effects Institute, 1995. adults: results from the AHSMOG study. Environ Health 72. Tockman MS. Other host factors and lung cancer susceptibility.
22. IARC. IARC Monographs on the evaluation of carcinogenic risks Perspect 106(12):813-823 (1998). In: Epidemiology of Lung Cancer (Samet JM, ed). New
to humans. Vol 46: Diesel and Gasoline Engine Exhausts and 48. Barbone F, Bovenzi M, Cavalleri F, Stanta G. Air pollution and York:Marcel Dekker, 1994;397-412.
Some Nitroarenes. Lyon:lnternational Agency for Research on lung cancer in Trieste, Italy. Am J Epidemiol 141(12):1161-1169 73. Raizenne M, Neas LM, Damokosh Al, Dockery DW, Spengler
Cancer, 1989. (1995). JD, Koutrakis P, Ware JH, Speizer FE. Health effects of acid
23. Lipsett M, Campleman S. Occupational exposure to diesel 49. Nyberg F, Gustavsson P, Jarup L, Bellander T, Berglind N, aerosols on North American children: pulmonary function.
exhaust and lung cancer: a meta-analysis. Am J Public Health Jakobsson R, Pershagen G. Urban air pollution and lung cancer Environ Health Perspect 104(5):506-514 (1996).
89:1009-1017 (1999). in Stockholm. Epidemiology (in press). 74. Peters JM, Avol E, Navidi W, London SJ, Gauderman WJ,
24. Health Effects Institute. Research Priorities for Mobile Air Tox- 50. Schulte PA, Perera FP, eds. Molecular Epidemiology: Principles Lurmann F, Linn WS, Margolis H, Rappaport E, Gong H Jr, et al.
ics. HEI Communication No 2. Cambridge, MA:Health Effects and Practices. New York:Academic Press, 1993. A study of twelve Southern California communities with differ-
Institute, 1993. 51. Motykiewicz G, Perera FP, Santella RM, Hemminki K, Seemayer ing levels and types of air pollution. I: Prevalence of respiratory
25. IARC. IARC Monographs on the Evaluation of Carcinogenic NH, Chorazy M. Assessment of cancer hazard from environmen- morbidity. Am J Respir Crit Care Med 159:760-767 (1999).
Risks to Humans. Vol 62: Wood Dust and Formaldehyde. tal pollution in Silesia. Toxicol Lett 88(1-3):169-73 (1996). 75. Peters JM, Avol E, Gauderman WJ, Linn WS, Navidi W, London
Lyon:lnternational Agency for Research on Cancer, 1995. 52. Perera FP, Hemminki K, Gryzbowska E, Motykiewicz G, Michalska SJ, Margolis H, Rappaport E, Vora H, Gong H Jr, Thomas D. A
26. Speizer FE, Samet JM. Air pollution and lung cancer. In: J, Santella RM, Young TL, Dickey C, Brandt-Rauf P, DeVivo I, et study of twelve Southern California communities with differing
Epidemiology of Lung Cancer (Samet JM, ed). New York:Marcel al. Molecular and genetic damage in humans from environmental levels and types of air pollution. Il: Effects on pulmonary func-
Dekker, 1994;131-146. pollution in Poland. Nature 360(6401):256-258 (1992). tion. Am J Respir Crit Care Med 159:768-775 (1999).
27. Doll R, Peto R. The Causes of Cancer. Oxford:Oxford University 53. Perera FP, Mooney LA, Dickey CP, Santella RM, Bell D, Blaner 76. Van de Lende R, Kok TJ, Reig RP, Quanjer PH, Schouten JP, Orie
Press, 1981. W, Tang D, Whyatt RM. Molecular epidemiology in environ- G. Decreases in VC and FEV1 with time: indicators for the effect
28. Dean G, Lee PN, Todd GF, Wicken AJ. Report on a Second mental carcinogenesis. Environ Health Perspect 104(suppl of smoking and air pollution. Bull Eur Physiopathol Respir
Retrospective Mortality Study in Northeast England: Parts 1 and 3):441-443 (1996). 17:775-792 (1981).
2. London:Tobacco Research Council, 1978. 54. Hemminki K, Dickey C, Karlsson S, Bell D, Hsu Y, Tsai WY, 77. Detels R, Tashkin DP, Sayre JW, Rokaw SN, Massey FJ,
29. International Agency for Research on Cancer (IARC). Cancer in Mooney LA, Savela K, Perera FP. Aromatic DNA adducts in Coulson AH, Wegman DH. The UCLA population studies of
Five Continents. Vol VIl. Lyons, France:lnternational Agency for foundry workers in relation to exposure, life style and CYPlA1 CORD. X: A cohort study of changes in respiratory function
Research on Cancer, 1997. and glutathione transferase Ml genotype. Carcinogenesis associated with chronic exposure to SO,, NO,, and hydrocar-
30. Kadafar K, Freedman LS, Goodall CR, Tukey JW. Urbanicity- 18(2):345-350 (1997). bons. Am J Public Health 81:350-359 (1991).
related trends in lung cancer mortality in US counties: white 55. Perera FP, Dickey C, Santella R, O'Neill JP, Albertini RJ, Ottman 78. Lubin JH, Gail MH. On power and sample size for studying fea-
females and white males, 1970-1987. Int J Epidemiol R, Tsai WY, Mooney LA, Savela K, Hemminki K. Carcinogen- tures of the relative odds of disease. Am J Epidemiol
25(5):918-932 (1996). DNA adducts and gene mutation in foundry workers with low- 131(3(:552-566 (1990).
31. Hannigan MP, Cass GR, Penman BW, Crespi CR, Lafleur AL, level exposure to polycyclic aromatic hydrocarbons. 79. Henderson BE, Gordon RJ, Menck H, Soohoo J, Martin SP, Pike
Busby WF Jr, Thilly WG. Human cell mutagens in Los Angeles Carcinogenesis 15(12):2905-2910 11994). MC. Lung cancer and air pollution in Southcentral Los Angeles
air. Environ Sci Technology 31(2):438-447 (1997). 56. Merlo F, Andreassen A, Weston A, Pan CF, Haugen A, Valerio F, County. Am J Epidemiol 101(6):477-488 (1975).
32. Pedersen DU, Durant JL, Penman BW, Crespi CL, Hemond HF, Reggiardo G, Fontana V, Garte S, Puntoni R, et al. Urinary excre- 80. Buffler PA, Cooper SP, Stinnett S, Contant C, Shirts S, Hardy RJ,
Lafleur AL, Cass GR. Seasonal and spatial variations in human tion of 1-hydroxypyrene as a marker for exposure to urban air Agu V, Gehan B, Burau K. Air pollution and lung cancer mortal-
cell mutagenicity of respirable airborne particles in the north- levels of polycyclic aromatic hydrocarbons. Cancer Epidemiol ity in Harris County, Texas, 1979-1981. Am J Epidemiol
eastern United States. Environ Sci Technology Biomarkers Prev 7(2):147-155 (1998). 128:683-699 (1988).
33(24):4407-4415 (1999). 57. Nielsen PS, de Pater N, Okkels H, Autrup H. Environmental air 81. Pike MC, Jing JS, Rosario IP, Henderson BE, Menck HR.
33. Pershagen G, Elinder CG, Bolander AM. Mortality in a region pollution and DNA adducts in Copenhaen bus drivers-effect of Occupation: `explanation" of an apparent air pollution related
surrounding an arsenic smelting plant. Environ Health Perspect GSTM1 and NAT2 genotypes on adduct levels. Carcinogenesis localized excess of lung cancer in los angeles county. In: Energy
19:133-137 (1977). 17(5):1021-1027 (1996). and Health (Breslow N, Whitemore A, eds). Philadelphia:SIAM,
34. WHO. Emissions of Heavy Metal and PAH Compounds from 58. Smith KR, Liu Y. Indoor air pollution in developing countries. In: 1979:3-16.
Municipal Solid Waste Incinerators: Control Technology and Epidemiology of Lung Cancer (Samet JM, ed). New York:Marcel 82. Vena JE. Air pollution as a risk factor in lung cancer. Am J
Health Effects. Report on a WHO meeting, Florence. Dekker, 1996:151-179. Epidemiol 116:42-56 (1982).
Copenhagen:World Health Organization, 1988. 59. Smith KR. Fuel combustion, air pollution exposure, and health: 83. Jedrychowski W, Becher H, Wahhrendorf J, Basa-Cierpialek Z.
35. Kogevinas M, Pearce N, Susser E, Boffetta P, eds. Social the situation in developing countries. Ann Rev Energy Environ A case-control study of lung cancer with special reference to
Inequalities in Cancer. No 138. Lyon:lnternational Agency for 18:529-566 (1993). the effect of air pollution in Poland. J Epidemiol Commun
Research on Cancer, 1997. 60. Kenworthy JR. Automobile dependence in Bangkok: an interna- Health 44:114-120 (1990).
36. Pershagen G. Air pollution and cancer. In: Complex Mixtures tional comparison with implications for planning policies and 84. Katsouyanni K, Karakatsani A, Messari I, Touloumi G, Hatzakis
and Cancer Risk (Pershagen G, Vainio H, Sorsa M, McMichael air pollution. In: Health at the Crossroads: Transport Policy and A, Kalandidi A, Trichopoulos D. Air pollution and cause specific
AJ, eds). Lyon:lnternational Agency for Research on Cancer, Urban Health (Fletcher T, McMichael AJ, eds). New York:John mortality in Athens. J Epidemiol Commun Health 44:321-324
1990;240-251. Wiley and Sons, 1997;215-234. (1990).
37. Archer VE. Air pollution and fatal lung disease in three Utah 61. WHO. Global Estimates for Health Situation Assessment and 85. Ozkaynak H, Schatz A, Thurston G. Relationship between
counties. Arch Environ Health 45:325-334 (1990). Projections. HST/90.2. Geneva:World Health Organization, 1990. aerosol extinction coefficient derived from airport visual range
38. Elliott P, Shaddick G, Kleinschmidt I, Jolley D, Walls P, 62. Xu ZY, Blot WJ, Xiao HP, Wu A, Feng WP, Stone BJ, Jie S, and alternative measures of airborn particle mass. Air Pollut
Beresford J, Grundy C. Cancer incidence near municipal solid Ershow AG, Henderson BE, Fraumeni JF. Smoking, air pollution, Control Assoc J 35:1 176-1185)(1985).
waste incinerators in Great Britain. Br J Cancer 73:702-710 and the high rates of lung cancer is Shenyang, China. JNCI
(1996). 8:1800-1806 (1989).

750 Environmental Health Perspectives * Vol 108, Supplement4 * August 2000