D26-03 The Association Between Traffic-Related PM2.5 Exposure and Lung Cancer Incidence: A Mixed Effects Approach

Abstract Rationale Ambient fine particulate matter (PM2.5) is a well-established risk factor for lung cancer. Different sources of PM2.5 have varying chemical compositions, which have profoundly different effects on human health. Proxies for traffic-related air pollution have been strongly associated with lung cancer incidence, but direct modeling of traffic-related PM2.5 and its relationship to lung cancer remains limited. Using a novel source-specific PM2.5 model, we examined the association between traffic-specific PM2.5 and county-level lung cancer incidence in the nationally representative Surveillance, Epidemiology, and End Results (SEER) database. Methods We linked annual county-level lung cancer incidence data from the SEER database (2011-2019) with county-level estimates of traffic and non-traffic PM2.5 exposure using a nationally validated hybrid satellite-land-use regression model. Counties with 10,000 residents were excluded. Our primary exposure was the 3-year prior mean (pre-3-year) value for source-specific PM2.5, and effects were estimated using within-county variation. We fit mixed-effects Poisson regression models with county random intercepts and year fixed effects, estimating incidence rate ratios (IRRs) per µg/m³ increase in PM2.5 with county level clustered bootstrap standard errors. Models were adjusted for county level covariates including index year PM2.5, smoking prevalence, age over 65, percentage rural, race, median income, high ozone days, and EPA radon group. A distributed lag linear model (DLM) was also performed to assess the time-varying effects of PM2.5 exposure on cancer incidence. Results A total of 512 counties with a total annual average population of 84,853,726 were included. In the unadjusted models, traffic-specific PM2.5 was associated with a borderline significant increased incidence rate of lung cancer (pre-3-year IRR 1.02, 95% CI 1.00-1.04, p 0.06, Fig 1a), while non-traffic PM2.5 showed a slightly negative relationship with cancer incidence (pre-3-year IRR 0.99, 95% CI 0.98-0.99, p 0.01). In the adjusted models, traffic-specific PM2.5 exposure was associated with a significant increase in cancer rates (pre-3-year IRR 1.03, 95% CI 1.01-1.05, p 0.01. Fig 1a) while non-traffic PM2.5 was associated with a decreased cancer incidence (pre-3-year IRR 0.99, 95% CI 0.98-0.99, p 0.01). DLM demonstrated that PM2.5 exposure had the strongest association in the years closest to cancer diagnosis, with progressive weakening at longer lags (Fig 1b and 1c). Conclusions Using our source-specific model of PM2.5, traffic-related PM2.5 was associated with an increased incidence of lung cancer, while non-traffic air pollution had a negative association. This finding supports source-specific pathophysiologic effects of PM2.5 and suggests distinct relationships to cancer pathogenesis. This abstract is funded by: Stony Wold-Herbert Foundation, NYU CTSA grant from the NCATS, NIH (UL1 TR001445)