Lung cancer clusters and air pollution exposure in Wayne County, Michigan: an autoregressive distributed lag model.

Background: Lung cancer is the most common cancer killer in the world. Although smoking is the biggest risk factor, it is increasingly being discovered that environmental pollution is also associated with the development of lung cancer. Previous studies have discovered the impact of environmental pollution on lung cancer, but it has been difficult to determine the duration of time after exposure in which the risk is greatest. Our goals were to investigate the association of specific air pollutants with lung cancer and to determine the lag time at which the incidence peaks following exposure. Methods: The Michigan Cancer Surveillance Program (MCSP) was queried from 1985 to 2018 for every case of lung cancer that occurred in Wayne County during that period. Data was obtained from the United States Environmental Protection Agency (EPA) for multiple pollutants from 1980 to 2021. Clusters of lung cancer in the county were detected by using SaTScan. Air pollution levels inside and outside the clustered areas were measured. The lagged relationship between pollution exposure and health outcome inside and outside the clustered areas were examined by using autoregressive distributed lag (ARDL) models, controlling for patient demographic features and meteorological conditions (MCs). Results: Three separate lung cancer clusters encompassing 26 zip codes were detected with a relative risk (RR) larger than 1. These three clusters were all located within urban and industrial areas and experienced higher air pollution than non-cluster areas. After controlling for MCs and sociodemographic variables, the optimal lag dramatically decreased to a shorter responsive period (2-3 years) than previous clinical expectations (9-12 years). Additionally, ozone exposure levels demonstrated shorter lag times inside cluster areas (2 years) than outside cluster areas (3 years). Furthermore, sulfur dioxide levels were significantly associated with lung cancer incidence inside cluster areas, while no pollutant showed a significant relationship in non-cluster areas. Conclusions: Although future studies are required to validate these findings, the ARDL model may play a role in estimating a responsive window for lung cancer incidence. The lag time between exposure to environmental pollutants and development of lung cancer may be shorter than previously thought based on this population-based study. Future research should continue to model air pollution lag times and lung cancer in different settings.