Lower levels of episodic memory associated with long‐term air pollution exposures in older women: Differences in midlife versus late‐life exposures and effect modification by hormone treatment

Background Late‐life exposure to ambient air pollutants may accelerate episodic memory (EM) declines in older women, but very little is known about the midlife exposure effects. Animal data suggests that hormone therapy (HT) may protect female brain from neurotoxic insult. We examined the EM trajectories associated with air pollutants and whether the associations varied by HT, exposure time‐window or pollution sources. Method We studied 916 women from the Women’s Health Initiative Memory Study in Younger (WHIMS‐Y) Women. At age of 50‐54, postmenopausal women were randomly assigned to an average of 5 years of conjugated equine estrogen alone (E‐alone) if they had hysterectomy, its combination with medroxyprogesterone acetate (E+P) if without hysterectomy or matching placebos. From the WHIMS‐Y inception age of 62‐71, EM was assessed annually (2008‐16) using the telephone‐based California Verbal Learning Test. Three‐year average exposures at midlife (aged 50‐59) and late‐life (at inception) were derived from yearly exposures estimated by regionalized universal kriging models for PM 2.5 (particulate matter <2.5μm) and NO 2 (a proxy of traffic‐related air pollutants). We used linear mixed‐effect models to estimate the associations, adjusting for social‐demographic variables, lifestyles, HT, hysterectomy, cardiovascular risk factors and other clinical covariates. Result EM performance improved (0.60/year, p<0.001) over the study period. Higher PM 2.5 exposures in midlife, but not in late‐life, were associated with lower EM levels (β=‐0.49, per inter‐quartile‐range [IQR]=3.44 µg/m 3 , p=0.045). NO 2 exposures in midlife or late‐life were not associated with EM levels. Compared to the women assigned to placebo group, the observed association between midlife PM 2.5 and EM level differed by HT (interaction p=0.005). The putative adverse PM 2.5 effect was greater (β=‐1.29 per IQR; p=0.03) in the E‐alone group, with no statistically significant difference in the E+P group (p=0.16). A similar pattern of effect modification was found for NO 2 exposure (interaction p=0.09) in midlife, but not in late‐life. The slope of EM trajectory was not associated with either exposure, HT, or their interactions. Conclusion Higher midlife PM 2.5 exposure decrease the EM levels of older women, with stronger associations in those randomly assigned to E‐alone treatment, which did not support the protection of HT against air pollution neurotoxicity.