In this study, we investigated the effects of exposure to urban and industrial air pollution on lung cancer risk in an industrialized area of Northern Spain. Our findings support the hypothesis that air pollution might be a risk factor for lung cancer. Indeed, our analyses indicate excess of risk of lung cancer among residents in both urban and industrial areas, though estimates failed to attain statistical significance. However, separate analyses of the two main health areas targeted for study confirmed excess risk in the Gijon area, while results in Aviles were mainly negative.
Difficulties in assessing environmental exposure and its long-term effects have posed numerous methodological problems in epidemiologic studies, pertaining mainly to the use of aggregated data for exposure and the lack of information on relevant confounders. Pending the development of adequate biomarkers of exposure, some authors have argued that well designed case-control studies with improved methods for retrospective assessment of exposure to industrial pollution and potential confounding factors should be used [9, 19]. To our knowledge, this is the first attempt to assess the influence of environmental pollution on lung cancer in Spain using individual data. The study design guarantees the availability of information on lung cancer risk factors, such as smoking habit and occupational exposure, which can be controlled for, and enables case and control exposures to be individually classified.
Insofar as environmental exposure is concerned, our measures were based on the residential location of the participants, and, despite the fact that were only able to take the geographical coordinates of subjects' last-reported residence into account, our study population proved to be very stable, i.e., 88.9% of cases and 89.3% of controls had lived in their last-reported residence for more than 5 years, and 80.0% of cases and 80.1% of controls had lived there for more than 10 years. We repeated the analyses with this last subgroup and we found similar results (data not shown). This variable affords relevant advantages for a case-control study, in that it cannot be expected to be influenced by recall bias. Moreover, the fact that we recruited incident cases also served to prevent possible changes of address associated with diagnosis of cancer. Hence, if there were any bias affecting proximity to pollution sources in relevant periods of life, our bias would be non-differential, causing an attenuation of the estimated effect.
Industrial pollution sources were identified using the EPER, which includes all industrial plants that have exceeded the reporting thresholds for one or more of the pollutants included in EU Decision 2000/479/EC. Subject to adequate validation of the geographical location of the data, this register has proved useful in ecological studies for ascertaining possible associations between residential proximity to such installations and mortality due to several cancers [17, 36–39]. To our knowledge, ours is the first case-control study to use publicly-available EPER data to analyze the effects of industrial pollution on cancer, and lung cancer in particular. Although this register includes quantitative data on pollutant emissions, the fact that this information was reported voluntarily raises doubts about its reliability. As a result, we preferred to use distance to pollutant sources as a proxy of population exposure.
Inevitably, the use of hospital-based controls is a potential limitation. In our case, the hospitals where the cases were recruited were reference centers for all patients requiring hospitalization. Our controls were referred to these hospitals owing to the presence of acute health conditions thought to be unrelated to lung cancer risk factors. The geographic distribution of the control population likely reflects population density in the health areas studied. Although there is always a chance of recall bias being present, due to the fact that information on confounding variables was obtained retrospectively, the estimators obtained for the most important of these -tobacco exposure and occupation- were nevertheless in line with the literature.
Exposure to industrial pollution was defined by the distance to the nearest industrial facility. It would have been of interest to analyze subjects' proximity to every industrial facility but, when we tried to perform this type of analysis, two additional problems arose: first, few individuals lived in the vicinity of each industrial installation, thereby severely limiting the statistical power; and second, proximity among industries led the exposure areas of several facilities to overlap, thus rendering individual interpretation of results difficult. We decided to use the strategy proposed by Barbone, and proceeded to define areas of exposure based on the geographic distribution of our controls .
Exposure to urban air pollution has been associated with increased lung cancer risk. It is well established that urban and outdoor air contains known and suspected human carcinogens. Urban air contains benzo[α]pyrene, benzene, and 1,3-butadiene, together with carbon-based particles onto which carcinogens may be adsorbed, oxidants such as ozone and nitrogen dioxide, and sulfur and nitrogen oxides in particle form . Outdoor air, particularly in densely populated urban environments, contains inorganic particulates (arsenic, asbestos, chromium and nickel), radionuclides (210Pb, 212Pb and 222Rn), and gaseous and particulate organic species (benzene, benzo[α]pyrene and benzene-soluble organics) . These substances are present as components of complex mixtures proceeding basically from combustion of fossil fuels for power generation or transportation. In this respect, Gijon is the most heavily populated city in Asturias, with 262,470 inhabitants (24.3% of the region's total population) , and it ranks among the most polluted cities in Spain. SO2, NO2, and, in particular, PM10 levels, monitored since 2000, repeatedly exceed the range set for air quality standards by European Union Directive 99/30/EC .
Our findings for industrial pollution are consistent with previous studies, including two reviews [9, 19], the former of which  reported in 1990 that most ecological studies showed an increased risk of lung cancer among populations living near non-ferrous smelters and a variety of other heavy industrial types. They noted, however, that few studies controlled for potential confounders, such as smoking or occupation. In 2001, Benedetti  reviewed 10 case-control studies: while seven of these reported an association between lung cancer risk and residential proximity to smelters, complex industrial areas, or localized sources of industrial emissions, three found little evidence of such an association. More recently, several studies have also observed associations between risk of lung cancer and: prolonged residence close to heavy industry ; residence within a 2-kilometer radius of a petrochemical plant in Brindisi (Italy) ; and residence in an urban area near a coke oven plant in Northern Italy .
Our results show that excess risks associated with urban and industrial pollution were concentrated in the Gijon health area, though in the case of industrial pollution the excess risk did not prove statistically significant. This area includes seven EPER industrial installations, four of which (two metal industries, a combustion installation, and a cement plant) could directly affect the general population due to their proximity to the city center. Several studies have reported excess lung cancer risks in the vicinity of these types of facilities. Metal industries located near the city center of Gijon include a steel foundry and an aluminum smelter, both associated with lung cancer risk in the literature. Indeed, steel founding is one of the industries classified by the International Agency for Research on Cancer (IARC) as implying a carcinogenic risk to humans , and several studies have reported an increased risk of lung cancer. These include a nested case-control study of Asturian iron and steel foundry workers  and a recent review of cohort studies conducted on workers exposed to PAHs . Similarly, aluminum production was classified by the IARC as a group 1 carcinogen , and several studies have reported excess risk of lung cancer associated with exposure to substances released by aluminum smelters [45–47]. With respect to fossil fuel-fired electric power plants, it is well established that these emit known or suspected carcinogens , with several studies having observed high concentrations of heavy metals in areas exposed to pollution from coal-fired power installations [15, 49], and a recent ecological study undertaken in Spain having reported excess lung cancer mortality among the population residing near Spanish combustion installations included in the EPER . Finally, insofar as cement plants are concerned, Fano observed a significant excess risk of lung cancer among people living in the proximity of a cement plant , an ecological study conducted in Lithuania documented excess risk of lung cancer among male cement workers , and a recent IARC multicenter case-control study on occupation reported an elevated lung cancer risk among men involved in the cement industry . Emissions from these industries include known or suspected carcinogens, such as arsenic, benzene, cadmium, chromium, dioxins, dichloromethane, lead, and nickel.
The Aviles health area contains nine EPER industrial installations (six metal industries, a glass installation, and two chemical plants), all of which are relatively far from the city center and, by extension, from the residence of most of the study population. Yet, the industries that are present in this area, mainly metal production and processing installations, have been associated with lung cancer risk in the literature. Nevertheless, the lack of association observed for the Aviles hospital area might also be due to lower statistical power, since we had only 176 cases and matched controls for this area.
Analyses by histologic type should be approached with care, owing to the small number of cases found in most categories of exposure. However, our results suggest that, while industrial air pollution may play a more specific role in the etiology of adenocarcinoma and small cell carcinoma, urban air pollution is associated with non-small cell lung cancer, whether squamous cell carcinoma or adenocarcinoma. The literature on histologic types of lung cancer and air pollution is limited. Barbone observed that air pollution was a moderate risk factor for certain histologic types of lung cancer in Trieste, Italy . Urban air pollution appears to increase the risk of small cell and large cell carcinoma, while the effects of industrial air pollution vary with the industrial process. In this connection, Barbone et al., found excess risk of adenocarcinoma in the shipyard section and increased risk of all histologic types in the incinerator section. In an earlier study conducted in China, Xu et al.,  found that the association with outdoor air pollution was stronger with squamous and oat cell cancer and adenocarcinoma. In occupational studies, Lubin and Blot  showed that both cigarette smoking and occupational exposure have stronger associations with squamous and small cell cancers than with adenocarcinoma. However, occupational exposure to asbestos appears to be more strongly associated with adenocarcinoma than with other types of lung cancer .
It is remarkable that the data on dispersion of industrial pollution emission may provide useful clues for evaluating results, but these are not available for the installations analyzed. Similarly, dispersion of the carcinogens present in air pollution is critically dependent on prevailing winds, with wind roses being one of the most useful tools for describing wind features. In Asturias, wind is a little-known climatic element owing to the small number of meteorological monitoring stations present in the region. The most remarkable and important fact is the pronounced seasonal nature of the wind, which makes it difficult to draw conclusions based on observation of prevailing winds in the area studied.