43: Monitoring and phenotyping rapid cystic fibrosis disease progression using community characteristics and environmental exposures

Background: The extent to which rapid CF disease progression is predicted by community characteristics and environmental exposures (geomarkers) is unknown. We sought to predict and phenotype rapid lung function decline using individual-level geomarkers.
Methods: We conducted a longitudinal cohort study (N = 33,972, ≥6 years old) of the CF Foundation Patient Registry (2003–2017). Geomarkers were ambient air pollution concentrations and hazard indices from the Environmental Protection Agency’s Environmental Justice Screening and Mapping Tool; land usage information from the Multi-Resolution Land Characteristics Consortium; indices of community material deprivation and crime, each linked to 5-digit zip codes. Novel longitudinal modeling with penalized variable selection was used to predict FEV1 decline with demographic/clinical characteristics and novel geomarkers as covariates. Covariate adjusted sparse functional principal component analysis was used to cluster pediatric patient-level FEV1 trajectories (aged 6–21). The first principal component score from multivariate geomarkers served as a covariate.
Results: In the overall population, established demographic/clinical predictors of rapid FEV1 decline were selected, including smoking (by the individual or exposure in primary residence). Modulator use corresponded to less decline. Selected geomarker-based risk factors included elevated exposure to PM2.5 and diesel particulate matter, total crime, and deprivation indices. Pediatric phenotypes of rapid decline corresponded to early, middle, and late timing of rapid decline. The first principal component from geomarker analysis had strong positive loadings for diesel particulate matter, air toxics respiratory hazard index, traffic proximity and volume, extent of impervious space, and a strong negative loading for extent of greenspace, representing a proxy of negative environment exposure. Early rapid decliners resided in areas with higher crime index (mean ± SD) were (89.3 ± 57.5), followed by middle (86.4 ± 59.8), then late (83.8 ± 59.7) decliners (all comparisons P < 0.05). Having rapid decline earlier associated with living near longer secondary roadways (early: 63593 ± 78297, middle: 59581 ± 73941, late: 55735 ± 68208; all comparisons P < 0.05). Elevated negative environmental exposure level linked to increased FEV1 trajectories in younger ages but decreased in older ages.
Conclusion: Accounting for neighborhood total crime, deprivation, and air pollution improves accuracy to predict rapid decline. Pediatric patients with an earlier rapid decline phenotype are more likely to reside in areas with more air pollution, more impervious spaces, increased crime, less tree cover, and limited greenspace. Negative environmental exposure more severely affects lung function during late adolescence/early adulthood compared to early childhood.