Association of Low-Level Ozone with Cognitive Decline in Older Adults

Abstract

Increasing evidence points to an association of airborne pollutant exposure with respiratory, cardiovascular, and neurological pathology. We examined whether or not ground-level ozone or fine particulate matter ≤ 2.5 μm in diameter (PM2.5) was associated with accelerated cognitive decline. Using repeated measures mixed regression modeling, we analyzed cognitive performance of a geographically diverse sampling of individuals from the National Alzheimer's Coordinating Center between 2004-2008. Ambient air concentrations of ozone and PM2.5 were established using a space-time Hierarchical Bayesian Model that statistically merged air monitor data and modeled air quality estimates. We then compared the ambient regional concentrations of ozone and PM2.5 with the rate of cognitive decline in residents within those regions. Increased levels of ozone correlated with an increased rate of cognitive decline, following adjustment for key individual and community-level risk factors. Furthermore, individuals harboring one or more APOE4 alleles exhibited a faster rate of cognitive decline. The deleterious association of ozone was confined to individuals with normal cognition who eventually became cognitively impaired as opposed to those who entered the study with baseline impairment. In contrast to ozone, we did not observe any correlation between ambient PM2.5 and cognitive decline at regulatory limits set by the Environmental Protection Agency. Our findings suggest that prolonged exposure to ground-level ozone may accelerate cognitive decline during the initial stages of dementia development.

Keywords: APOE4; Air pollution; Alzheimer’s disease; ozone; particulate matter.

Fig. 1

Geospatial processing of gridded ozone predictions. A) Spatial coverage of HBM air quality data and the 3-digit boundary map. B) 12 km × 12 km resolution of the 2001–2006 HBM data zoomed in on a Northern California region. C) 36 km × 36 km resolution of the 2007-2008 HBM data zoomed in on a Northern California region. D) Interpolated result of the HB modeled ozone data at the 36-km grid extent for a Northern California region. E) Interpolated result of the HB modeled 2004 ozone data for the contiguous United States colored by lowest (green) to highest (red) levels. Ozone levels are highest in the Southwestern states due to distinct meteorological and topological conditions, most notably the proximity to a large body of water [46].

Fig. 2

Age distribution of all NACC participants at baseline. Opaque bars show age of cognitively impaired participants (BLMMSE ≤ 24) in comparison to all (transparent bars).

Fig. 3

MMSE assessment of NACC participants at baseline.

Fig. 4

Distribution of participants according to 3-digit residential ZIP codes. ADC sites are depicted by red circles. Created using Weave v.1.9.38 [47].

Fig. 5

Annual average concentrations of ozone in ppb (A) and PM_2.5 in μg/m³ (B) during the 2004–2008 observation period as calculated in Methods.

Fig. 6

Performance on the MMSE and CDR-SB for all participants by pollutant. Cognitive performance trajectories based on MMSE for the three tertiles of ozone (A) and PM_2.5 (B). Cognitive performance trajectories based on CDR-SB for the three tertiles of ozone (C, the y-axis is reversed) and PM_2.5 (D, the y-axis is reversed). Cognitive performance trajectories based on MMSE for interaction of APOE and ozone (E) and APOE and PM_2.5 (F). Values represent the intercept plus the slope of time in years modified by each pollutant category (mean ± std. error), *p < 0.05 versus reference (i.e., highest ozone and PM_2.5 in A-D; Highest ozone and PM_2.5/no E4 in E, F).