Association of environmental exposome and cognitive function among older adults with and without preclinical Alzheimer's disease

Abstract

INTRODUCTION

Sociodemographic predictors are established risk factors for Alzheimer's disease (AD), a leading cause of cognitive decline. This study examines the influence of environmental exposures like walkability, green space, and light pollution among older adults.

METHODS

Cross‐sectional analysis examined associations between walkability, green space, tree canopy, light pollution, and cognitive functioning among participants with and without preclinical AD based on cerebrospinal fluid (CSF) and/or positron emission tomography (PET) biomarkers. Generalized additive mixed models (GAMMs) assessed their impact on cognitive performance, controlling for sex and race as categorical covariates, with age and education included as continuous variables modeled using smooth functions.

RESULTS

Walkability, green space, and tree canopy were significantly associated with better cognitive performance, while light pollution negatively impacted cognition. Biomarker positivity was not a significant predictor.

DISCUSSION

Our findings highlight the crucial role of environmental exposures on cognition, suggesting that urban design and environmental quality influence cognitive trajectories in older adults, regardless of biomarker status.

Highlights

• Walkability, green space, and tree canopy improve cognitive performance.

• Light pollution is linked to lower cognitive function in older adults.

• Biomarker positivity was not a significant predictor of cognitive outcomes.

• Urban design plays a key role in cognitive health and Alzheimer's risk.

1. BACKGROUND

Alzheimer's disease (AD) is a neurodegenerative condition with distinct clinical stages, beginning with a long preclinical stage without cognitive impairments, then progressing to mild cognitive impairment, and then to a prodromal stage. AD's projected incidence will increase to 13 million by 2050, highlighting a public health crisis, particularly given the rising life expectancy in the United States. ¹ Age is a well‐established risk factor for AD. ² Environmental and lifestyle factors are increasingly recognized for their role in cognitive decline, particularly in the preclinical stage. ³ Research suggests that nearly half of individual differences in AD risk may be attributed to environmental influences, including urbanicity versus rurality and lower education levels, which have been linked to earlier onset in individuals with genetic risk factors. ⁴ , ⁵

The concept of the AD exposome—an encompassing framework assessing lifetime environmental exposures that affect AD risk—has emerged in recent studies. ⁶ Originally applied in cancer research, ⁷ , ⁸ the exposome concept extends to AD, providing insights into how multiple factors across the lifespan contribute to disease vulnerability. ⁹ Environmental exposures such as air pollution, walkability, light, noise, and greenspace have significant impacts on cognitive health, particularly in vulnerable populations. ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ For example, air pollution, especially fine particulate matter (PM_2.5), has been linked to accelerated brain aging, reduced white matter volume, and increased amyloid‐beta (Aβ) deposition, all hallmarks of AD. ¹⁰ , ¹¹ Similarly, long‐term exposure to traffic‐related noise contributes to cognitive impairment through sleep disturbances and stress, exacerbating amyloid plaque formation. ¹² A systematic review and dose‐response meta‐analysis by Meng et al. ¹³ found that for every 57 dB increase in noise exposure, the relative risk (RR) of developing dementia rose by 47%, with a 25 dB increase resulting in a 16% rise in dementia risk (RR: 1.16, 95% confidence interval [CI]: 1.12–1.20). This association was consistent across subtypes of dementia, including AD, vascular dementia, and non‐AD dementia.

Greenspace and tree canopy offer protective effects for cognitive function. Studies show that neighborhoods with more greenspace are associated with slower cognitive decline and greater resilience to neurodegenerative stressors. ¹⁴ , ¹⁵ These benefits may be mediated by increased physical activity, social engagement, and reduced exposure to harmful pollutants in greener environments. ¹⁶ , ¹⁷ Conversely, the lack of walkable environments and greenspaces, especially in socioeconomically disadvantaged areas, can increase AD risk and accelerate cognitive decline. ¹⁸ , ¹⁹ Chronic exposure to artificial light at night (ALAN) is also increasingly recognized as a risk factor for neurodegenerative diseases like AD. ALAN disrupts circadian rhythms, which are essential for regulating Aβ and tau proteins, both key to AD pathology. Light pollution reduces melatonin production, impairs sleep quality, and leads to oxidative stress, accelerating cognitive decline. ²⁰ As a result, disrupted or insufficient sleep can lead to the buildup of Aβ plaques in the brain, as the glymphatic system or the brain's natural clearance processes for these proteins are less effective during poor‐quality or shortened sleep interval. ²¹

Demographic factors, such as age, education, sex, and race, also play a critical role in moderating the impact of environmental exposures on cognitive health. Older adults with lower education levels show heightened vulnerability to the negative effects of pollution and noise on cognitive function. ²² , ²³ Furthermore, populations self‐identified as Black, Hispanic/Latino, Indigenous, and Asian are more likely to reside in neighborhoods with higher levels of environmental pollutants, which may contribute to the disproportionately higher rates of cognitive impairment and dementia. ²⁴ , ²⁵ , ²⁶ , ²⁷

Preclinical AD is the asymptomatic 15–20 years characterized by accretion of amyloid and tangles in the brain without manifesting cognitive or functional symptoms. ²⁸ Cerebrospinal fluid (CSF), and positron emission tomography (PET) imaging are markers for identifying preclinical AD. ²⁹ Emerging evidence suggests that early identification of preclinical AD can provide critical insights into how environmental factors can impact progression. ³⁰ , ³¹ This study investigated the additive effects of environmental factors on cognitive functioning among older adults with and without preclinical AD.

2. METHODS

2.1. Participants

Participants were enrolled in prospective, longitudinal observational studies focused on aging and dementia as a part of the Driving Real‐world In‐Vehicle Evaluation System (DRIVES) Project at Washington University School of Medicine. The cohort was comprised of community‐dwelling residents from both St. Louis City and St. Louis County in the metropolitan area. To be eligible, participants had to (1) be age 65 or older; (2) have a valid driver's license; (3) be cognitively normal, based on the Clinical Dementia Rating score of 0 (CDR = 0); and (4) be willing to complete biomarker testing via CSF collection and/or PET imaging. The Washington University Human Research Protection Office approved the procedures. Participants were excluded if they were diagnosed with mild cognitive impairment or dementia (Clinical Dementia Rating scale [CDR] score > 0), had significant neurological or psychiatric illness, had incomplete biomarker or cognitive data, or were unwilling or unable to complete the imaging or cerebrospinal fluid (CSF) procedures.

2.2. Cognitive assessments

Participants complete an annual battery of neuropsychological tests assessing various cognitive processes. A Preclinical Alzheimer's Cognitive Composite (PACC) score was computed. ³² The PACC score included the free recall test from the Selective Reminding Test, containing the Free Recall (SRTFREE), which measured episodic memory, and Trail Making Test Parts A and B, which measured processing attention, sequencing, and inhibition, and Animal fluency, which measured verbal fluency. The absolute z‐scores were computed for each measure, and then the scores were averaged into a composite.

2.3. Biomarker measurement

Participants completed CSF and PET biomarkers every 2–3 years. CSF was collected as described in previous studies. ³³ Analytes like Aβ₄₂, Aβ₄₀, p‐Tau/Aβ₄₂, and t‐Tau/Aβ₄₂ levels were measured using automated electrochemiluminescence immunoassay (Lumipulse G1200, Fujirebio, Tokyo, Japan). Although tau‐related biomarkers (t‐Tau and p‐Tau) were measured and recorded, only Aβ₄₂/Aβ₄₀ ratio was used to classify preclinical AD status in this study. A participant was considered biomarker‐positive if the Aβ₄₂/Aβ₄₀ ratio was less than 0.0673, a threshold validated to maximize concordance with PET imaging. ³⁴ PET amyloid imaging was completed with either [¹¹C]Pittsburgh Compound‐B (PiB) or florbetapir (F‐AV‐45) tracer. The binding potentials and standardized uptake value ratios (SUVRs) were computed utilizing the designated post‐injection windows of 30–60 min for PiB and 40–70 min for F‐AV‐45, applying the cerebellar gray matter as the reference region. The global amyloid burden was represented by averaging regions known to be sensitive to AD pathology. The amyloid burden was derived from the mean cortical standardized uptake value ratio (MCSUVR), which incorporated partial volume correction through the regional spread function (RSF). Preclinical AD was identified as PET‐positive if the MCSUVR RSF exceeded 1.42 for PiB ³⁵ and 1.19 for F‐AV‐45 tracers. ³⁶

Magnetic resonance imaging (MRI) scans were used for co‐registration and anatomical alignment of PET images to ensure accurate localization and quantification of amyloid burden. All structural MR scans (T1w and fluid‐attenuated inversion recovery [FLAIR]) were acquired on BioGraph mMR PET‐MR 3T and Siemens TIM Trio 3 T MRI scanners using a protocol designed to match the Alzheimer's Disease Neuroimaging Initiative (ADNI)‐2 MRI protocol. ³⁷ T1w images were processed with FreeSurfer 5.3 and resampled to 1 × 1 × 1 mm resolution for volumetric segmentation and cortical reconstruction. ³⁸ Thickness and gray matter volumes for 68 cortical regions and volumes for 12 subcortical gray matter structures in left and right hemispheres were derived after quality control of FreeSurfer output through visual inspection and manual editing of the cortical and subcortical segmentation output when necessary. All MRI scans were processed using FreeSurfer version 5.3 (http://surfer.nmr.mgh.harvard.edu/). FreeSurfer performed cortical reconstruction and segmentation of T1‐weighted brain images. These steps include correcting for head motion, identifying subcortical and deep gray matter structures, and normalizing image intensity. Each brain was then aligned to a standard atlas based on individual folding patterns and divided into regions based on the structure of the brain's gyri and sulci. Detailed descriptions of these methods are available in previous publications. ³⁹

RESEARCH IN CONTEXT

1. Systematic review: We reviewed literature from traditional databases (e.g., PubMed) to examine environmental factors and demographic variables associated with cognitive decline and Alzheimer's disease (AD). We focused on studies exploring the impact of neighborhood environments (e.g., walkability, green space, light, noise) and their interactions with demographic factors (age, education, race, and biomarkers) in preclinical AD.

2. Interpretation: Our findings align with previous studies showing that environmental exposures, such as poor walkability and limited green space, are linked to accelerated cognitive decline in older adults with preclinical AD. Demographic variables further modulate this relationship, with socially disadvantaged groups exhibiting greater vulnerability. This study contributes to growing evidence on the role of environmental and social determinants in AD risk and progression.

3. Future directions: Future research should explore the mechanisms underlying the interaction between environmental and demographic factors and their influence on AD pathology. Longitudinal studies could assess the long‐term effects of these exposures and inform targeted interventions.

In cases where PET and CSF biomarkers were available, participants were classified as biomarker‐positive if either modality exceeded its threshold.

2.4. Environmental exposome

To assess the environmental factors contributing to cognitive decline, we analyzed several key exposures, including noise, light pollution, walkability, green space, and tree canopy, all of which have been implicated in increasing AD risk. The participant's home address (latitude, longitude) was used as the geographic reference for the exposure data. Given the longitudinal nature of the study, the most recent time point (time interval) was used for participants.

2.4.1. Noise exposure

Noise data were obtained from the National Transportation Noise Exposure Map, ⁴⁰ which overlays the Bureau of Transportation Statistics' noise map with 5‐year population estimates from the American Community Survey. The original noise was calculated at receptor locations. The noise receptors are defined by a uniform grid with a resolution of 98.4 feet (30 m). Each receptor is modeled at a height of 4.92 feet (1.5 m) above ground level. Noise levels are adjusted to account for ground effects and free‐field divergence differences between the source reference and receptor locations. We aggregated their estimates at the census tract level to determine the proportion of the population exposed to transportation‐related noise levels above 60 decibels (dB), a threshold known to contribute to adverse health outcomes. This percentage was used to quantify noise pollution exposure for participants' neighborhoods.

2.4.2. Light pollution

Data on artificial light exposure were derived from the New World Atlas of Artificial Night Sky Brightness (2016), ⁴¹ which utilizes newly available low‐light imaging data from the Visible Infrared Imaging Radiometer Suite (VIIRS) Day/Night Band (DNB) sensor aboard the Suomi National Polar‐orbiting Partnership (NPP) satellite. The DNB provides nightly global coverage with a swath width of approximately 3000 km, and each pixel has a spatial resolution of 742 m. The atlas compiles composite data from 6 months of VIIRS observations, processed by the Earth Observation Group (EOG) at National Oceanic and Atmospheric Administration (NOAA). We extracted pixel‐level radiance values corresponding to participants' addresses to quantify their exposure to artificial light pollution.

2.4.3. Walkability

Neighborhood walkability was calculated using a kernel density approach, assessing the density of walkable roads within a 750 m radius of each participant's residence. Data on road networks was obtained from the NAVTEQ mapping database, which includes detailed road attributes. Roads with a speed limit of over 54 miles per hour or less than 8 miles per hour and roads in airports, cemeteries, and parking lots were excluded to focus on pedestrian‐friendly environments. This measure reflects the availability of safe and accessible walking paths in participants' neighborhoods, consistent with previous studies on walkability. ⁴² , ⁴³

2.4.4. Green space and tree canopy

Green space and tree canopy exposure were calculated as the percentage of green space and tree cover within a 100 m buffer along walkable roads. These data were sourced from EnviroAtlas, ⁴⁴ which provides detailed land cover information. We intersected these data with walkable road networks to capture the natural environment surrounding participants’ residences, considering both green space and tree canopy as important buffers against environmental stressors. By incorporating these environmental metrics, we aim to capture the cumulative effects of the exposome on cognitive health, providing a comprehensive assessment of participants' neighborhood environments.

2.5. Model specification

A generalized additive mixed model (GAMM) was employed to explore the non‐linear relationships between PACC score and various environmental and demographic variables. GAMMs are particularly well‐suited for this analysis because they allow for flexible modeling of non‐linear relationships while accounting for random effects, making them ideal for data structures that involve repeated measures. GAMM or its variants have been widely used in AD research to assess cognitive performance, disease progression, and the impact of environmental and genetic factors. ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸

Table 1 presents all the variables used in the GAMM. The key predictor variables were categorized into two main groups: Environmental Factors and Demographic Factors. Time since the start of the study was included as a predictor variable to account for potential temporal effects. Additionally, random intercepts for participant IDs were specified to account for within‐subject correlation resulting from repeated measures collected from the same individuals over time. The model was fitted using the gamm4 package in R, which integrates mixed‐effects modeling with Generalized Additive Models (GAMs). Smooth terms were used to capture potential non‐linear effects for the environmental predictors (walkability, light, noise, green space, and tree canopy). These smooth terms were estimated using a thin plate regression spline with a specified number of knots (k = 15), providing enough flexibility to model non‐linear patterns without overfitting the data. Age, education, and time in the study were also modeled using smooth terms to capture their potential non‐linear effects. All variables modeled as smooth terms were standardized using z‐scores before model fitting to facilitate interpretability and comparability of effect sizes. The model's parametric terms included linear effects for the categorical predictors such as sex, race, and biomarker status. The generalized cross‐validation (GCV) score, effective degrees of freedom (edf), and estimated scale parameter were used to evaluate the model's fit. This approach allowed for the capture of complex relationships between predictors and cognitive decline while accounting for individual variability through random effects. The inclusion of smooth functions enabled the modeling of nonlinear effects, which is crucial for understanding the nuanced influence of environmental exposures on cognitive health outcomes.

TABLE 1.

Variables used in the GAMM.

Variable type	Acronyms	Description
Outcome variable	Preclinical Alzheimer's Cognitive Composite	Cognitive functioning is measured by a composite Z‐score, representing the cognitive performance of a participant at one time.
Environmental factors	Walkability	A measure of the density of walkable roads in each participant's neighborhood.
	Light	Quantified using satellite imagery to represent the level of artificial light exposure in participants’ residential areas.
	Noise	Estimated using national transportation noise data to capture ambient noise levels.
	Green space	The percentage of green space in the participants’ neighborhood.
	Tree canopy	The percentage of tree coverage in the participants’ neighborhood.
Demographic factors	Baseline age	The participant's age at the start of the study.
	Education	The number of years of education completed by the participant.
	Sex	A binary variable comparing male to female.
	Race	Participants’ self‐reported racial identity.
Endogenous factors	Preclinical AD	Indicates whether the participant is in the preclinical stage based on cut‐off for CSF or PET biomarker.
	Progression to preclinical AD	Indicates whether there was a change in CSF or PET biomarkers for the participant.
Temporal factor	Time in the study	Time since the beginning of the study to account for potential temporal effects.
Random effects	Identification	Random intercepts specified to account for the correlation within repeated measures from the same individual.

Abbreviations: AD, Alzheimer's disease; CSF, cerebrospinal fluid; GAMM, generalized additive mixed model; PET, positron emission tomography.

3. RESULTS

The study included 272 participants with a mean age of 73.17 (SD = 5.03) years at baseline. Among the participants, 126 were male and 146 were female. In terms of racial composition, 236 participants self‐identified as White, and 36 self‐identified as Black or African American. Education levels varied, with an average of 16.54 years of education (SD = 2.44) across the cohort. A portion of the study cohort exhibits biomarkers indicative of underlying AD pathology. Specifically, 63 patients were classified as positive for CSF biomarkers, and 57 patients were positive on PET imaging. Notably, 39 patients showed positivity on both tests. Table 2 presents the baseline characteristics of the study sample.

TABLE 2.

Participant characteristics.

Variable	Mean (SD)	Range	N (%)
Baseline age (years)	73.17 (5.03)	57.2–90.9
Education (years)	16.54 (2.44)	12.0–24.0
Sex
Female			146 (53.7%)
Male			126 (46.3%)
Race
Black or African American			36 (13.2%)
Non‐Hispanic White			236 (86.8%)
Biomarker status (CSF or PET)
Negative			191 (70.2%)
Positive			81 (29.8%)
CSF positive			63
PET positive			57
Both positive			39

Abbreviations: CSF, cerebrospinal fluid; PET, positron emission tomography.

3.1. Model performance

The model explained approximately 23.89% of the variance in cognitive functioning based on the PACC score, with an adjusted R‐squared of 18.45%. The GCV score for the model was 0.129, and the estimated scale parameter was 0.120, indicating the model's fit to the data. Table 3 reported each parametric term's estimates, standard errors, t‐values, and p‐values. The edf and reference degrees of freedom (Ref.df) were reported for each smooth term (Table 4), along with the F‐statistics and p‐values to indicate the significance of the non‐linear effects. Figure 1 illustrates the model's predictive performance, comparing predicted and observed cognitive scores, with a reference line (y = x) indicating perfect agreement.

TABLE 3.

Coefficients of parametric terms

Term	Estimate	Std. error	t‐value	p‐value
(Intercept)	−0.005	0.084	−0.055	0.956
Preclinical AD (CSF or PET positive)	−0.024	0.031	−0.769	0.442
Male (sex)	−0.094	0.029	−3.186	0.001
White (race)	−0.005	0.044	−0.104	0.917
Participant ID (random effects)	0.000	0.000	0.810	0.418

Abbreviations: AD, Alzheimer's disease; CSF, cerebrospinal fluid; PET, positron emission tomography.

TABLE 4.

Approximate significance of smooth terms

Term	edf	Ref. df	F	p‐value
Smooth term: Time in study	1.510	1.850	0.412	0.580
Smooth term: Walkability	10.874	12.470	4.262	<0.001
Smooth term: Baseline age	1.000	1.000	0.019	0.890
Smooth term: Education (years)	1.799	2.258	2.121	0.099
Smooth term: Light exposure	12.184	13.352	3.447	<0.001
Smooth term: Noise exposure	6.462	7.723	1.905	0.059
Smooth term: Green space	11.559	12.964	4.580	<0.001
Smooth term: Tree canopy	10.805	12.396	3.472	<0.001

Abbreviations: edf, effective degrees of freedom; Ref.df, reference degrees of freedom.

FIGURE 1.

Comparison of predicted versus observed cognitive scores with reference line.

3.2. Parametric terms

The model's parametric terms indicated that the effect of sex was statistically significant, with males showing a lower cognitive score compared to females (Estimate = −0.094, SE = 0.031, t = −3.186, p = 0.001). However, the effects of the presence of either CSF or PET positivity in these biomarkers were not statistically significant (Estimate = −0.024, SE = 0.084, t = −0.055, p = 0.956). Race also did not significantly impact cognitive scores (Estimate = −0.005, SE = 0.044, t = −0.104, p = 0.917), and the random effect of participant ID (edf = 0.000, SE = 0.044, t = −0.104, p = 0.418) were not statistically significant.

3.3. Smooth terms

The model's smooth terms provided insight into the non‐linear relationships between cognitive scores and the continuous predictors. Significant non‐linear effects were observed for several environmental factors. Walkability was significantly associated with cognitive scores (edf = 10.874, F = 4.262, p < 0.001). The smooth function indicated a complex relationship, suggesting that cognitive scores varied non‐linearly with changes in walkability. Light exposure also showed a significant non‐linear effect on cognitive scores (edf = 12.184, F = 3.447, p < 0.001). The relationship between light exposure and cognitive scores appeared to be non‐linear, indicating that different levels of light exposure might have varying impacts on cognitive function. Green space was another significant predictor, with a non‐linear effect on cognitive scores (edf = 11.559, F = 4.580, p < 0.001). The model suggested that cognitive scores were influenced by the amount of greenspace in the participants' living environments. Tree canopy also demonstrated a significant non‐linear effect (edf = 10.805, F = 3.472, p < 0.001), indicating that the amount of tree cover in the living environment plays a role in cognitive outcomes. Conversely, the non‐linear effects of time since start (edf = 1.51, F = 0.412, p = 0.580), age at baseline (edf = 1.00, F = 0.019, p = 0.890), education (edf = 1.799, F = 2.121, p = 0.099), and noise (edf = 6.50, F = 1.871, p = 0.063) were not statistically significant.

3.4. Smooth effects on cognitive scores

Figure 2 illustrates the effects of smooth terms on cognitive scores. The plots demonstrate the non‐linear relationships between cognitive performance and variables such as walkability, light exposure, green space, tree canopy, age, and education, allowing us to visualize how these factors influence cognitive outcomes across a range of values. Walkability showed a significant non‐linear association with cognitive scores. The plot suggests that cognitive performance initially increases with higher walkability, but after a certain threshold, the relationship becomes more complex, indicating diminishing returns or even negative effects at extreme levels of walkability. Light exposure exhibited a similarly non‐linear relationship, with cognitive scores declining as light pollution increased. The plot shows that moderate levels of light pollution have less of an impact, but higher levels lead to sharper declines in cognitive function, supporting previous findings on the negative effects of light pollution on circadian rhythms and cognition. Green space and tree canopy also showed significant non‐linear effects. For green space, cognitive scores improve with increasing exposure, although the relationship plateaus at higher levels, suggesting that while green environments are beneficial, additional green space beyond a certain point may not provide additional cognitive benefits. Tree canopy demonstrated a similar trend, where moderate coverage was associated with better cognitive outcomes, but very high tree canopy density might reduce exposure to natural light, impacting cognitive performance. Age and education had less pronounced smooth effects compared to environmental factors. While older age is generally associated with lower cognitive scores, the relationship appears relatively linear, with a gradual decline as age increases. Education, a well‐known protective factor against cognitive decline, shows a subtle positive effect on cognitive scores, particularly for individuals with higher educational attainment.

FIGURE 2.

Smooth effects of environmental and demographic variables on cognitive scores.

4. DISCUSSIONS

4.1. Summary of findings

Environmental and demographic factors play significant roles in the development and progression of cognitive decline, particularly in the preclinical stage of AD. Studies have consistently identified advanced age as the strongest risk factor for AD, with prevalence increasing significantly beyond the age of 75. ⁴⁹ Education is found to be another crucial determinant, as lower educational attainment is associated with an increased risk of cognitive decline and dementia due to reduced cognitive reserve, which may limit resilience to neurodegenerative changes. ⁵⁰ Racial and ethnic disparities also influence AD risk, with Black and Hispanic populations experiencing a 1.5‐ to 2‐fold higher risk compared to non‐Hispanic White populations, largely driven by socioeconomic and healthcare access disparities. ⁵¹ Additionally, women have a higher likelihood of developing AD, which may be attributed to hormonal, genetic, and lifestyle factors. ⁵²

Beyond demographic factors, environmental influences have garnered increasing attention. Chronic exposure to air pollution, heavy metals, and pesticides has been linked to neuroinflammation and increased AD pathology. ⁵³ Social and environmental complexity, including engagement in mentally stimulating activities and living in diverse environments, has been suggested as protective against cognitive decline, potentially delaying the onset of AD symptoms. ⁵⁴ Lifestyle factors such as physical activity, nutrition, and social engagement also contribute to cognitive resilience. ⁵⁵ Although these were not assessed in our study, they represent important avenues for future research when considered alongside environmental exposures.

Emerging research in preclinical AD has focused on identifying biomarkers such as Aβ and tau proteinopathies, which can predict disease progression before clinical symptoms appear. These biomarkers, combined with an understanding of environmental and demographic risk factors, may facilitate early interventions aimed at slowing cognitive decline and mitigating AD‐related pathology.

Our model, which incorporated both parametric and smooth terms, examined the complex interactions between demographic, environmental, and cognitive factors. Parametric terms such as sex significantly impacted cognitive scores, with males showing lower cognitive performance than females. However, CSF/PET biomarker status and race did not have a significant effect. Non‐linear relationships were observed for several environmental factors, with walkability, light exposure, green space, and tree canopy all significantly influencing cognitive scores. Meanwhile, factors such as noise, time since baseline, age, and education did not have significant non‐linear effects.

4.2. Interpretation of key results

The significant association between male sex and lower cognitive scores aligns with previous studies highlighting sex differences in cognitive aging and AD progression. ⁵⁶ While postmenopausal women are often considered at higher risk due to hormonal changes, our findings suggest that males exhibit worse cognitive outcomes, potentially due to differences in cognitive reserve, lifestyle factors, or genetic predispositions. ⁵⁷ Some studies indicate that men are exposed to more environmental stressors, such as pollution and noise, due to occupation, which could contribute to their heightened risk of cognitive decline. ⁵⁸ Further research is needed to explore these mechanisms in greater detail.

Interestingly, positive or changed CSF/PET biomarkers did not significantly impact cognitive scores in this cognitively normal, preclinical cohort. These findings challenge conventional assumptions that these biomarkers strongly predict early cognitive decline. ⁵⁹ One possible explanation is that environmental factors, such as walkability or green space, play a more dominant role in shaping cognitive outcomes in young and midlife, providing a reserve for early‐stage AD. ⁶⁰ This finding further highlights the need for a multifactorial approach to assessing AD risk that incorporates both biological and environmental influences.

One of the most striking findings is the non‐linear relationship between walkability and cognitive outcomes. Prior research has consistently linked walkable environments with enhanced cognitive health, likely due to increased physical activity, social engagement, and reduced stress. ⁶¹ However, our results suggest that these benefits may vary across different levels of exposure, potentially moderated by additional environmental factors such as noise and air pollution. Similarly, light exposure exhibited a non‐linear effect, supporting emerging evidence that excessive nighttime light exposure disrupts circadian rhythms and heightens neurodegenerative risk. ⁶² This reinforces concerns about light pollution's impact on cognitive function and the need for more targeted urban planning strategies. ⁶³ Green space and tree canopy, often considered protective environmental factors, also displayed non‐linear relationships with cognitive outcomes. While moderate exposure to green environments is linked to lower stress, increased physical activity, and improved mental health, excessive tree canopy density might limit natural light exposure, influencing mood and circadian regulation. ⁶⁴ These findings contribute to an evolving understanding of how environmental complexity shapes cognitive health. Contrary to expectations, noise exposure did not significantly influence cognitive scores, despite prior studies linking high noise levels to cognitive impairment through mechanisms such as sleep disruption and chronic stress. ⁶⁵ This may be due to our study population's specific noise exposure levels or potential mitigating factors such as neighborhood walkability. Future studies should explore more granular noise exposure measures, including duration and intensity, to fully understand its impact on cognition.

Age and education, two well‐established predictors of cognitive function, also did not exhibit significant non‐linear effects. This could be due to the relatively homogeneous nature of our preclinical AD cohort, wherein the influence of these variables might become more apparent as the disease progresses. Additionally, the presence of strong environmental predictors in our model may have attenuated the direct impact of age and education on cognitive outcomes.

4.3. Implications for the field

Our findings highlight several key implications for understanding cognitive decline in preclinical AD. The lack of significance for CSF/PET biomarkers suggests that environmental factors play a crucial, mediating role in shaping cognitive health, even in individuals with biological markers of AD. This underscores the importance of expanding our focus beyond biomarkers to include broader exposome assessments when evaluating AD risk.

The significant effects of walkability, light, green space, and tree canopy emphasize the role of urban design in cognitive health. Policy‐makers and urban planners should consider these environmental factors when designing interventions aimed at reducing cognitive decline, particularly in vulnerable populations. Improving access to green spaces, increasing walkability, and mitigating light pollution could be valuable strategies for promoting cognitive resilience.

Another important finding relates to sex differences in environmental impact. The significant effect of sex on cognitive scores, combined with the non‐significance of race, suggests that sex may moderate the relationship between environmental exposures and cognitive health. Future research should aim to disentangle the biological and social drivers of these differences to inform targeted interventions that consider sex‐specific susceptibilities to environmental risk factors.

This study also highlights the necessity of a multidisciplinary approach to understanding AD risk and progression. While previous research has primarily focused on genetic and lifestyle factors, our results suggest that built and natural environments should be integrated into AD risk models. Recognizing the interplay between environmental, demographic, and biological factors will lead to more comprehensive strategies for prevention and early intervention. The absence of Indigenous, Asian, and Hispanic participants in our sample limits the generalizability of our findings to these populations, who may face distinct environmental exposures and social determinants of health. Future research should aim to include underrepresented populations to better assess environmental influences on cognitive health across diverse racial and ethnic groups. Although our sample included participants from St. Louis City and County, representing a range of urban and suburban neighborhoods, the findings may not fully generalize to more rural areas or regions with different demographic or environmental profiles. Future research should expand to broader geographic settings to assess the consistency of these associations.

4.4. Conclusion

This study highlights the significant role that environmental exposures, particularly walkability, light pollution, and green space, play in cognitive functioning in older adults with and without preclinical AD. While traditional biological markers like CSF/PET positivity were not significant predictors in this model, the complex interactions between environmental factors and cognitive outcomes underscore the need for a holistic approach to AD risk assessment.

Given the cross‐sectional nature of this study, future longitudinal research is necessary to better understand the causal relationships between environmental exposures and cognitive decline over time. Expanding the range of environmental variables considered in future models, such as air pollution, heat islands, and access to health services, could further refine our understanding of the exposome's role in AD progression. Moreover, investigating the interactions between environmental and genetic risk factors may reveal new opportunities for mitigating AD risk through environmental interventions.

The findings related to walkability, light, and green space suggest that environmental modifications could serve as viable strategies for slowing cognitive decline in preclinical AD populations. Future research should explore the effectiveness of such interventions, potentially through randomized controlled trials, to determine their impact on delaying the onset of cognitive symptoms and improving the quality of life for at‐risk populations. By integrating environmental, biological, and social determinants of health, researchers and policy‐makers can develop more effective approaches to AD prevention and care.

CONFLICT OF INTEREST STATEMENT

All authors declare no conflicts related to this work. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

All participants provided written informed consent, and all study procedures were approved (202010214, 202003209) by the Washington University Human Research Protection Office.

Supporting information

Supporting Information

Li K, Shacham E, Brown D, et al. Association of environmental exposome and cognitive function among older adults with and without preclinical Alzheimer's disease. Alzheimer's Dement. 2025;21:e70373. 10.1002/alz.70373