Differential DNA methylation in blood as potential mediator of the association between ambient PM2.5and cerebrospinal fluid biomarkers of Alzheimer’s disease among a cognitively normal population-based cohort

Differential DNA methylation in blood as potential mediator of the association between ambient PM2.5 and cerebrospinal fluid biomarkers of Alzheimer’s disease among a cognitively normal population-based cohort | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Differential DNA methylation in blood as potential mediator of the association between ambient PM 2.5 and cerebrospinal fluid biomarkers of Alzheimer’s disease among a cognitively normal population-based cohort View ORCID Profile Tszshan Ma , Jiaqi Liu , Donghai Liang , Stefanie Ebelt , Kyle Steenland , Allan I. Levey , James J. Lah , Aliza P. Wingo , Thomas S. Wingo , View ORCID Profile Anke Hüls doi: https://doi.org/10.1101/2025.04.15.25325910 Tszshan Ma 2 Ganagarosa Department of Environmental Health, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Tszshan Ma Jiaqi Liu 3 Department of Psychiatry, University of California , Davis, Sacramento, CA USA , 95816 Find this author on Google Scholar Find this author on PubMed Search for this author on this site Donghai Liang 1 Department of Epidemiology, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 2 Ganagarosa Department of Environmental Health, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 Find this author on Google Scholar Find this author on PubMed Search for this author on this site Stefanie Ebelt 1 Department of Epidemiology, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 2 Ganagarosa Department of Environmental Health, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 Find this author on Google Scholar Find this author on PubMed Search for this author on this site Kyle Steenland 1 Department of Epidemiology, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 2 Ganagarosa Department of Environmental Health, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 Find this author on Google Scholar Find this author on PubMed Search for this author on this site Allan I. Levey 4 Goizueta Alzheimer’s Disease Research Center, Emory University School of Medicine , Atlanta, GA USA , 30322 5 Department of Neurology, Emory University School of Medicine , Atlanta, GA USA , 30322 Find this author on Google Scholar Find this author on PubMed Search for this author on this site James J. Lah 4 Goizueta Alzheimer’s Disease Research Center, Emory University School of Medicine , Atlanta, GA USA , 30322 5 Department of Neurology, Emory University School of Medicine , Atlanta, GA USA , 30322 Find this author on Google Scholar Find this author on PubMed Search for this author on this site Aliza P. Wingo 6 Division of Mental Health , Northern California VA, Sacramento, CA USA , 95816 3 Department of Psychiatry, University of California , Davis, Sacramento, CA USA , 95816 Find this author on Google Scholar Find this author on PubMed Search for this author on this site Thomas S. Wingo 7 Department of Neurology, University of California , Davis, Sacramento, CA USA , 95816 8 Alzheimer’s Disease Research Center, University of California , Davis, Sacramento, USA , 95816 Find this author on Google Scholar Find this author on PubMed Search for this author on this site Anke Hüls 1 Department of Epidemiology, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 2 Ganagarosa Department of Environmental Health, Rollins School of Public Health, Emory University , Atlanta, GA, USA , 30322 Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Anke Hüls For correspondence: anke.huels{at}emory.edu Abstract Full Text Info/History Metrics Data/Code Preview PDF Abstract Introduction Fine particulate matter (PM 2.5 ) is a known risk factor for Alzheimer’s disease (AD), with emerging evidence linking PM 2.5 exposure to cerebrospinal fluid (CSF) biomarkers in pre-clinical stages. However, the role of DNA methylation (DNAm) as potential mediator in this relationship among cognitively normal individuals remains largely unexplored. Methods In 535 cognitively normal individuals, we assessed genome-wide blood DNAm, CSF Aβ 42 concentrations, and residential PM 2.5 exposure in the year preceding blood collection. Multi-stage comprehensive mediation analyses were conducted. Results Nine CpG sites mediated the PM 2.5 –Aβ42 association, with significant natural indirect effects (NIEs) for eight CpGs, mediating 14–43% of the effect. The joint NIE for all nine CpGs was -0.115 (95% CI: -0.215, -0.101) per 1 ug/m 3 increase in PM 2.5 exposure. Six CpGs are annotated to genes implicated in neuroinflammatory pathways. Discussion Our findings suggest that differential DNAm, particularly in neuroinflammation-related genes, mediates PM 2.5 toxicity in AD’s pre-clinical stage. 1 Background Alzheimer’s disease (AD), the most common form of dementia, is characterized by the progressive accumulation of beta-amyloid (Aβ) plaques and neurofibrillary tangles of hyperphosphorylated tau in the brain, leading to neuronal cell death and cognitive impairment[ 1 ]. With aging populations, AD has become a growing public health challenge[ 2 ]. The majority of AD cases occur in individuals over 65 and result from a combination of genetic predispositions, such as the apolipoprotein E ( APOE ) ε4 allele, and environmental exposures[ 3 – 5 ]. Increasing epidemiological studies have associated air pollution, particularly fine particulate matter (PM 2.5 ), with increased risk of AD and related dementia[ 6 , 7 ], and biomarkers of AD pathology, including Aβ and tau proteins[ 8 – 10 ]. However, the underlying biological mechanisms mediating these associations remain incompletely understood. Epigenetic regulation, particularly DNA methylation (DNAm), plays a critical role in neuronal function and cognitive health[ 11 ]. Growing research implicates DNAm in AD pathogenesis[ 12 ], with studies reporting DNAm alterations in key genes such as amyloid-beta precursor protein ( APP) [ 13 – 15 ], presenilin 1 ( PSEN1) [ 16 ], and APOE [ 13 , 17 ] in postmortem brain samples. Epigenome-wide association studies (EWASs) have identified additional AD-associated loci, including TMEM59 , which is involved in APP post-translational processing[ 18 ]; DUSP22 , which may regulate tau phosphorylation[ 19 ]; ANK1 , which plays a role in neuronal membrane compartmentalization[ 20 , 21 ]; and the HOXA gene cluster, which contributes to embryonic development and neuroprotective function[ 22 , 23 ]. Recently, meta-analyses of EWAS datasets validated previously identified genes and uncovered new differentially methylated positions related to inflammatory responses, DNA repair, and cell cycle regulation[ 24 , 25 ]. Beyond its role in AD, DNAm is also influenced by environmental exposure such as PM 2.5 exposure. Studies have reported that long-term PM 2.5 exposure is linked to global DNA hypomethylation, potentially leading to genomic instability[ 26 , 27 ]. PM 2.5 exposure has also been associated with DNAm changes in inflammation-related genes [ 12 ], such as tumor necrosis factor ( TNF-a) and interleukin-6 ( IL6) , suggesting a role in promoting systemic inflammation[ 28 , 29 ]. Similarly, PM 2.5 -induced DNAm changes in oxidative stress-related genes, such as glutathione S transferase ( GST ) and inducible nitric oxide synthase (iNOS) , may contribute to cellular damage and impaired detoxification pathways[ 28 , 30 ]. EWASs have further identified PM 2.5 -related methylation alterations in loci involved in immune pathways, oxidative stress, and neuroinflammation, reinforcing the plausibility of epigenetic mediation in PM 2.5 -induced neurotoxicity[ 31 – 33 ]. However, formal mediation analyses assessing the causal role of DNAm in linking PM 2.5 exposure to AD remain scarce. We previously conducted the first human study linking traffic-related PM 2.5 , DNAm, and AD neuropathy using postmortem human brain samples from the Emory Goizueta AD Research Center (ADRC) brain bank[ 34 ]. Our findings suggested that differential DNAm in genes related to neuroinflammation, such as AKT1S1 (Proline-rich AKT1 substrate 1), may mediate PM 2.5 ’s effect on AD neuropathological markers in brain[ 34 ]. While DNAm in brain tissue offers the most direct evidence of AD-related epigenetic changes[ 12 ], blood-based analyses allow for earlier detection in cognitively normal individuals, enabling investigation of preclinical AD biomarkers and air pollution exposure in living populations[ 35 ]. Building on this foundation, the current study explored the relationship between ambient PM 2.5 exposure, blood DNAm, and the AD cerebrospinal fluid (CSF) biomarker Aβ 42 levels in cognitively normal participants from the Emory Healthy Brain Study (EHBS). We previously demonstrated that ambient PM 2.5 is significantly associated with decreased CSF Aβ 42 levels in this cohort[ 9 ]. To elucidate the underlying biological mechanisms—potentially involving epigenetic regulation and neuroinflammation—we examined whether differential DNAm in blood mediates the association between PM 2.5 and CSF Aβ42 levels. To test this hypothesis, we applied a multi-stage analytical pipeline integrating traditional single-mediator analysis[ 36 ], high-dimensional mediation analyses, including DACT, HIMA1, HIMA2, HDMA, and MedFix [ 37 – 41 ], and causal mediation analysis for validation[ 42 , 43 ]. Given the absence of a gold-standard approach for high-dimensional DNAm mediation analysis, this multi-method framework enhances the likelihood of identifying biologically meaningful CpG sites[ 44 ]. This study provides critical insights into the role of DNAm in mediating the effects of PM 2.5 exposure on the pre-clinical stage of AD, shedding light on potential mechanisms relevant for intervention and prevention efforts. 2 Methods 2.1 Study Population The study population for this analysis included participants in the Emory Healthy Brain Study (EHBS), a prospective research study focusing on the cognitive health of middle-aged and older adults. The EHBS is nested within the Emory Healthy Aging Study (EHAS) and includes participants 45−75 year of age who were free of cognitive impairment at enrollment from the Atlanta metropolitan region in the state of Georgia in the United States. More details on recruitment and eligibility can be found elsewhere[ 45 ]. The EHBS biennial study visits include neuropsychological tests, biospecimen collection (blood, CSF), cardiovascular measures, and brain imaging performed by trained clinical research staff[ 45 ]. Demographic characteristics of participants were collected with the online Healthy History Questionnaire (HHQ)[ 45 ]. The current cross-sectional analysis included data from the baseline visits conducted between 2016 and 2020. Of the 1,113 participants with CSF biomarker measurements, 536 also had genome-wide DNA methylation (DNAm) data available. The current analysis focuses on these 536 participants with both DNAm and CSF biomarker data. All participants provided informed consent, and the study received approval from the local ethics committee and the Emory University Institutional Review Board. 2.2 AD CSF Biomarker Concentrations CSF biospecimens were collected by EHBS research staff via lumbar puncture at enrollment. CSF Aβ 42 levels were quantified using the ElectroChemiLuminescense Immunoassay (ECLIA) Elecsys AD CSF portfolio on an automated Roche Diagnostics instrument (F. Hoffman-La Roche Ltd). The assay has measuring ranges of 200– 1,700 pg/mL for Aβ 42 . CSF Aβ 42 level was normalized as z-score and kept as a continuous outcome in our main analyses. 2.3 Genome-wide DNA methylation measurement Whole blood biospecimens were collected at enrollment and used for DNA extraction. DNAm measures were obtained with the MethylationEPIC BeadChips. The raw intensity files were transformed into a dataset that included beta values for each of the CpG sites, with values computed as the ratio of the methylated signal to the sum of the methylated and unmethylated signals, which ranged from 0 to 1 on a continuous scale. Pre-processing and statistics were done using R (v4.2.0). Following sample and probe filtering, 536 samples and 661,869 CpG sites remained for down-stream analyses. More details on the preprocessing and quality control have been published elsewhere[ 46 ]. The DNAm beta values were kept as a continuous mediator in our analyses. Whole blood cell type proportions (percentages of CD8+ T cells, CD4+ T cells, natural killer cells, B cells, monocytes, and neutrophils) was predicted using the most recent whole blood reference data set and the R package minfi [ 47 , 48 ]. 2.4 PM 2.5 Exposure Assessment High-resolution ambient PM 2.5 exposure data was obtained from the publicly available Socioeconomic Data and Application Center (SEDAC) air quality dataset for health-related applications[ 49 ]. The data consists of yearly average ambient PM 2.5 concentrations (in μg/m 3 ) estimated at a 1 km × 1 km resolution for the entire contiguous United States (2000–2016). The PM 2.5 predictions were derived using established spatiotemporal ensemble models that integrated three different machine learning algorithms (neural network, random forest, and gradient boosting machine) and included over one hundred predictors from satellite measurements, land-use terms, meteorological variables, and chemical transport model predictions[ 50 ]. The quality of the estimates was assessed using 10-fold cross-validation against monitoring measurement values from the Environmental Protection Agency (EPA) Air Quality System across the United States. The resulting R 2 values for national annual predictions were 0.89[ 50 ]. As our study participants all lived in the state of Georgia, we restricted the PM 2.5 estimates to Georgia. Participants’ geocoded residential addresses at their baseline visit were mapped using QGIS mapping software and we spatially matched geocoded residential addresses to the closest centroid of 1-km 2 PM 2.5 grids to assign annual exposures at the year for specimen collection[ 9 ]. 2.5 Covariates Individual-level demographic characteristics (self-reported gender, age, race/ethnicity, educational attainment, and body mass index (BMI)) were obtained from the online Health History Questionnaire[ 45 ]. Self-reported gender was a binary variable: female, male. Age as a continuous variable denotes the age at the year of visit for biospecimens collection. Because of the presence of multi-ancestral groups and small categories in self-reported race/ethnicity, we used a three-level race variable: White, Black/African American, and Other, as well as a binary ethnicity variable indicating Hispanic origin. Similarly, educational attainment was included as a three-level variable: less than a college degree, college degree, master’s degree or higher. Height and weight measurements were used to calculate BMI (weight in kilograms divided by height in square meters) as a continuous variable. We also included two area-level characteristics, Area Deprivation Index (ADI) and principal components of neighborhood deprivation, to denote participant’s neighborhood socioeconomic status (SES). The ADI is provided in national percentile rankings at the census block group level from 1 to 100, where 100 represents the most deprived neighborhood, and was calculated using census block group-level indicators and factor analysis to cluster indicators based on their ability to explain the variance between block groups[ 51 ]. The three principal components of neighborhood deprivation were calculated based on estimates from American Community Survey (ACS) census tract-level data, including 15 indicators of 6 socioeconomic domains (poverty/income, racial composition, education, employment, occupation, and housing properties) (see Li et al. 2022[ 52 ] for details). We assigned each participant’s ADI, and three principal components of neighborhood deprivation variables based on their residential address[ 9 ]. In addition to individual-level and area-level demographic covariates, we also include whole blood cell-type proportions as covariates due to their substantial impact on DNA methylation. The confounding structure was determined according to literature review and our previous studies[ 9 ], and is illustrated by directed acyclic graphs (DAGs) in Figure S1 . 2.6 Statistical Analysis Total Effect Analysis Before conducting the mediation analysis, we assessed the total effect of PM 2.5 exposure on CSF Aβ 42 concentrations without considering DNAm as a mediator. Previously, we found higher PM 2.5 exposure was associated with decreased CSF Aβ 42 levels among 1,113 participants in the EHBS cohort, suggesting an accumulation of amyloid plaques in the brain and an increased risk of developing AD [ 9 ]. We repeated this total effect analysis among the subset of 536 participants with DNAm data with the following linear regression model: Here, PM 2.5 is the continuous exposure variable, Aβ 42 is the continuous outcome as the CSF Aβ 42 z-score, and C denotes the adjustment set of covariates, including sex, age, race/ethnicity, educational attainment, BMI, ADI, three principal components of neighborhood deprivation, as shown in the DAG ( Figure S1A ). Mediation Effect Analysis Next, we aimed to identify blood DNAm patterns that potentially mediate the association between PM 2.5 exposure and decreased CSF Aβ 42 levels. Given the lack of consensus in the literature regarding best practices for conducting mediation analysis with high-dimensional mediators[ 53 ], we developed a comprehensive analysis pipeline similar to our previous work [ 34 , 54 , 55 ] ( Figure 1 ; see Figure S2 for additional details on each method). We began by conducting a traditional single-mediator mediation analysis using the product method to assess each CpG site individually and screen for CpG sites potentially associated with both PM 2.5 exposure and decreased CSF Aβ 42 levels[ 36 ]. Next, we applied five high-dimensional mediation analysis methods (DACT, HIMA1, HIMA2, HDMA, and MedFix) to identify noteworthy mediators among the CpG sites filtered in the first step [ 37 – 41 ]. These high-dimensional methods encompassed a diverse array of statistical approaches, increasing the likelihood of identifying relevant mediators[ 44 ]. Finally, we validated the noteworthy CpG sites identified in the second step by performing causal mediation analysis to estimate the natural indirect effects of each CpG site and the joint indirect effect of all noteworthy CpG sites [ 42 , 43 ]. All analyses were completed in R (v4.4.1). Download figure Open in new tab Figure 1. Graphical overview of the study design and the analytical pipeline used to identify and evaluate potential mediating CpG sites in the association between PM 2.5 (exposure) and CSF AD biomarker Aβ 42 concentrations (outcome). The pipeline incorporates single-mediator analysis, high-dimensional mediation analysis, and causal mediation analysis. Note: PM 2.5 , fine particulate matter; CSF, cerebral spinal fluid; AD, Alzheimer’s disease; Aβ 42 , beta-amyloid 42; C EM and C MO denote the covariate matrices for the mediator model and outcome models, respectively. Step 1: Traditional single-mediator mediation analysis We first assessed the CpG site “one-at-a-time” using the standard regression-based approach proposed by Baron and Kenny (1986)[ 36 ]. The following robust linear regression models were fitted for each CpG site: to estimate the mediating role of DNAm in the causal pathway between the exposure PM 4.5 and outcome Aβ 4 2 [ 56 ]. Model (2) represents the outcome model, while Model (3) represents the mediator model. Here, PM 4.5 is the continuous exposure variable, Aβ 42 is the continuous outcome as the CSF Aβ 42 z-score, and DNAm the continuous mediator as the DNAm beta value. Covariates in the adjustment sets C MO (for the outcome model) and C EM (for the mediator model) included sex, age, race/ethnicity, educational attainment, BMI, ADI, three principal components of neighborhood deprivation, and proportion of blood cells, as shown in the DAGs for the mediator and outcome models ( Figure S1B and S1C ). To identify the natural indirect effect (NIE) mediated through DNAm, the following assumptions need to be met[ 57 , 58 ]: (i) no unmeasured exposure-outcome confounding conditional on C ; (ii) no unmeasured mediator-outcome confounding conditional on ( C , PM 4.5 ); (iii) no unmeasured exposure-mediator confounding conditional on C ; (iv) no effect of exposure that confounds the mediator-outcome relationship; and (v) no exposure-mediator interaction on the outcome. Under these standard assumptions, the NIE is ⍺ a β m , the direct effect of exposure on outcome is β a , and the total effect is β a + ⍺ a β m . We fitted the above robust linear regression models for each CpG site using R packages MASS , lmtest , and sandwich . Only CpG sites with p-values <0.05 for both the exposure-mediator association ( ⍺ a ) and the mediator-outcome association ( β m ) were included in downstream analyses. Furthermore, we retained only CpG sites with negative indirect effects ( ⍺ a β m < 0) to align with previously observed direction of the total effect[ 9 ]. This pre-screen step strategically reduced dimensionality of the data, facilitating the feasibility of subsequent high-dimensional mediation analysis[ 54 , 55 ]. Step 2: High-dimensional mediation analysis We applied five high-dimension mediation analysis methods—DACT, HIMA1, HIMA2, HDMA, and MedFix—to identify noteworthy mediators among the filtered CpG sites. Since these methods are relatively novel, there is currently no established gold standard for high-dimensional mediation analysis[ 53 ]. DACT ( divide-aggregate composite null test ) by Liu et al. (2022) utilized the Efron empirical null framework to calculate a weighted sum of p-values obtained from the outcome and mediator models (Model 2 and 3) to test for the significance of mediators[ 37 ]. The remaining four methods (HIMA1, HIMA2, HDMA, and MedFix) extended Model (1) and (2) by fitting multiple mediators simultaneously using penalized regression for variable selection and effect estimation[ 44 ]. HIMA1 ( high-dimensional mediation analysis ) by Zhang et al. (2016)[ 38 ] employed a dimension reduction technique based on sure independence screening (SIS) using p-values from the single-mediator outcome model[ 59 ]. It then applied the minimax concave penalty (MCP)[ 60 ] for mediation effect estimation and joint significance testing, assuming a uniform null distribution for p-values[ 61 , 62 ]. HDMA ( high-dimensional mediation analysis ) by Gao et al. (2019)[ 40 ] developed along the lines of HIMA1 but replaced the MCP with the de-biased LASSO penalty[ 63 ] for mediation effects estimation. HIMA2, developed by Perera et al. (2022), was an extended version of HIMA1[ 39 ]. Like HDMA, HIMA2 also utilized the de-biased LASSO penalty for effect estimation but improved the SIS approach by ranking CpG sites based on the absolute magnitude of estimate indirect effect (αβ). Additionally, it implemented a p-value correction procedure that allowed a mixture of null distributions[ 64 ]. MedFix ( mediation analysis via fixed effect model ) by Zhang (2022) extended HIMA for settings involving multiple exposures and multiple mediators[ 41 ]. It employed adaptive LASSO for effect selection in both the outcome and mediator models[ 65 ]. Although MedFix was designed for multi-exposure scenarios, in our single-exposure setting, it functioned similarly to HDMA but used adaptive LASSO instead of debiased LASSO. We conducted DACT analysis using the R package DACT , while HIMA1, HDMA, and MedFix analyses were performed using the R package hdmed . HIMA2 analysis was carried out using the R package HIMA . In each separate high-dimensional mediation analysis, we classified a CpG site as “active” if its estimated mediation contribution was non-zero and its raw p-value was <0.05. Following the recommendation in Clark-Boucher et al. (2023), we considered “active” CpG sites identified by at least two methods as noteworthy mediators[ 44 ]. Step 3: Causal mediation analysis Noteworthy CpG sites were validated using causal mediation analysis performed with the R package CMAverse , which provided estimates of NIE, natural direct effect (NDE), total effect (TE), and proportion mediated (PM)[ 42 , 43 , 58 ]. We assessed mediators under two scenarios: (1) individually, allowing for exposure-mediator (E-M) interactions during effect estimation, and (2) jointly, considering all noteworthy CpG sites together while also allowing for E-M interactions. This multi-mediator approach relaxed the assumption that there were no mediator-outcome confounders affected by the exposure, allowing for more realistic modeling where one mediator might influence another within the causal pathway[ 66 ]. This approach enhanced the biological plausibility of the estimates by reflecting the interconnected nature of CpG sites within biological pathways. The 95% confidence intervals for effect estimates were constructed using bootstrapping, and two-sided p-values were reported. 2.7 Sensitivity analyses Given that CSF AD biomarker concentrations may differ by race[ 67 ], we repeated the causal mediation analysis for the identified noteworthy CpG sites, restricting the study population to White participants, who comprised 85.4% of our cohort (see Table S1 for descriptive characteristics of the White EHBS participants). 2.8 Secondary analyses To further support our findings, we conducted a blood-brain concordance analysis for CpG sites identified as noteworthy mediators in the high-dimensional mediation analyses (step two). We assessed the correlation between DNAm in blood and brain tissue using the Blood-Brain Epigenetic Concordance (BECon) tool[ 68 ] and the Gene Expression Omnibus Database [Accession code GSE111165]. Additionally, we performed a gene ontology (GO) functional enrichment analysis using the R package missMethyl [ 69 ], based on the broader set of CpG sites filtered from the standard single-mediator mediation analysis in step one, which were associated with PM 2.5 exposure and decreased CSF Aβ 42 levels. All CpG sites were annotated using the annotation data for the “IlluminaHumanMethylationEPIC” array[ 70 ]. For further functional insight into individual CpGs sites and their corresponding annotated genes, we searched previous studies using the EWAS catalog[ 71 ] and conducted literature searches on PubMed. 3 Results 3.1 Study Population Characteristics The analysis included 536 EHBS participants with available CSF AD biomarker data, whole blood DNAm and other relevant covariates ( Table 1 ). The average age of the study population was 61.8 years (standard deviation [SD] = 7.0), with 70.3% identifying as female. The majority of participants (85.4%) identified as White, 9.5% as Black/African American, and 2.8% as Hispanic. The average BMI was 25.5 kg/m 2 (SD = 3.7). Most participants, 449 (83.8%), completed college or higher education, and lived in less deprived neighborhoods compared to the national average (ADI: mean = 29.7, SD = 20.4). As shown by the 1-year average residential PM 2.5 exposure map ( Figure 2 ), participants residing in the southern areas of the Atlanta city experienced higher PM 2.5 levels compared to those living in northern communities. The mean of 1-year ambient PM 2.5 exposure was 9.53 μg/m 3 (interquartile range [IQR] = 0.79), slightly above the latest annual National Ambient Air Quality Standards (NAAQS) PM 2.5 threshold of 9 μg/m 3 [ 72 ]. CSF Aβ 42 concentrations displayed considerable variability within the study population (median = 1230.5 pg/mL, IQR = 640.8) ( Table 1 ). While CSF Aβ 42 concentrations followed an approximately normal distribution, there was a notable frequency peak at the highest levels, indicating that most participants had Aβ 42 concentrations within the normal range ( Figure S3 ). Download figure Open in new tab Figure 2. Geographic distribution of the study population and their residential PM 2.5 exposure concentrations in the year prior to specimen collection. Each dot represents a participant (n = 536) and their 1-year averaged ambient residential PM 2.5 exposure (in ug/m 3 ), categorized by quintiles. View this table: View inline View popup Download powerpoint Table 1. Baseline descriptive characteristics, PM 2.5 exposure and AD CSF biomarker outcomes for the EHBS study participants. 3.2 The total effect of PM 2.5 on CSF Aβ 42 concentration The total effects of PM 2.5 on CSF Aβ 42 concentration were estimated using linear regression adjusting for confounders including sex, age, race/ethnicity, educational attainment, BMI, ADI, and three principal components of neighborhood deprivation ( Table S2 ). A negative but statistically insignificant association was observed between 1-year average PM 2.5 exposure and CSF Aβ 42 , with each 1 μg/m 3 increase in PM 2.5 corresponding to a -0.074 (95% CI: -0.190, 0.042) decrease in CSF Aβ 42 z-score. When the analysis was expanded to include a larger sample of EHBS participants not limited to those with DNAm data (N = 1,113), the observed negative association was statistically significant (β = -0.102; 95% CI: -0.179, -0.024), as reported in our previous study[ 9 ]. 3.3 High-dimensional mediation analyses Standard single-mediator mediation analysis was conducted to assess the CpG site “one-at-a-time” and screen for potential mediators associated with PM 2.5 exposure and decreased CSF Aβ 42 levels. Of the 661,869 CpGs tested, there were 1,967 showed p-values < 0.05 for both exposure-mediator and mediator-outcome associations. Among these, 898 CpG sites demonstrated a negative natural indirect effect (NIE), consistent with the negative total effect observed between PM 2.5 exposure and CSF Aβ 42 concentrations ( Figure S4 ). These 898 CpGs were advanced to downstream high-dimensional mediation analyses using five methods: DACT, HIMA1, HIMA2, HDMA, and MedFix ( Figure 1 ). We classified a CpG site as “active” if its estimated mediation effect was non-zero and its raw p-value for significance testing was <0.05 in the high-dimensional mediation analyses. Among the 898 CpG sites analyzed, DACT identified 111 active sites, HIMA1 identified 13, HIMA2 identified 9, HDMA identified 5, and Medfix identified 4, yielding a total of 133 unique CpG sites ( Table S3 ). Nine CpG sites were identified as active by two of the methods and were considered noteworthy mediators in our analysis pipeline ( Table 2 ). No CpG site was identified as active by more than two methods. The estimated mediation effects for these nine sites were consistent across methods in both effect size and direction, all exhibiting negative effects in line with the expected direction of the total effect ( Table 2 ). Based on chromosomal annotation, six CpG sites were located within or near annotated genes: cg0053427[ SLC7A11 ], cg11676342 [ BEND7 ], cg17904786 [ GTF3A ], cg05532093 [ NFATC1 ], cg16891376 [ FLT3LG ], cg05352541 [ PITPNB ]. The remaining three CpG sites were intergenic: cg21682780, cg23129342, and cg10285526. View this table: View inline View popup Download powerpoint Table 2. Estimated mediation effects and the raw joint significant test p-value of noteworthy CpG sites identified by at least two high-dimensional mediation analysis methods ( DACT , HIMA1 , HIMA2 , HDMA , MedFix ) for the associations between PM 2.5 exposure and AD CSF biomarker Aβ 42 concentrations a . We further validated these nine CpG sites using causal mediation analysis to evaluate the significance of their NIEs under settings that accommodated exposure-mediator (E-M) interactions and accounted for potential mediator-mediator influence within the causal pathway[ 43 , 66 ]. When each CpG site was evaluated individually, eight of the nine demonstrated significant negative NIEs ( Figure 3 ), while E-M interaction effects were not significant for any CpG tested. The eight CpG sites with significant NIEs included cg21682780 [ intergenic ], cg00534274 [ SLC7A11 ], cg11676342 [ BEND7 ], cg17904786 [ GTF3A ], cg10285526[ intergenic ], cg05532093[ NFATC1 ], cg16891376 [ FLT3LG ], and cg05352541 [ PITPNB ]. All effect estimates reflect changes in the CSF Aβ 42 z-score per 1 ug/m 3 increase in PM 2.5 exposure. The NIE estimates (95% CI) for the eight significant CpG sites, ranged from -0.015 (- 0.036, -0.0008) for cg21682780 [ intergenic ] to -0.029 (-0.062, -0.0069) for cg05532093 [ NFATC1 ]. All NIEs were smaller than the natural direct effects (NDEs) but exhibited greater precision. When all nine CpG sites were considered simultaneously as multiple mediators, the estimated joint NIE increased to -0.115 (95% CI: -0.215, - 0.101) ( Figure 3 ), while E-M interaction effects were again not significant. Download figure Open in new tab Figure 3. Causal mediation analysis for the association between PM 2.5 (exposure) and AD CSF biomarker Aβ 42 concentrations (outcome, standardized as z-scores) using noteworthy CpG sites identified through high-dimensional mediation analysis methods. The figure presents estimates for the natural indirect effect (NIE), natural direct effect (NDE), total effect (TE), and proportion mediated (PM) for selected CpG sites, assessed both individually and jointly. Effect estimates represent the change in Aβ 42 z-score concentrations per 1ug/m 3 increase in PM 2.5 exposure. Significant indirect effects are marked with (*), while no significant direct or total effects were observed. 3.4 Sensitivity analyses To evaluate the robustness of our finding, we repeated the causal mediation effect of the nine noteworthy CpG sites, restricting the study population to White participants (N=458) ( Table S1, Figure S5 ). Seven CpG sites retained statistically significant NIEs, including cg21682780 [ intergenic ], cg11676342 [ BEND7 ], cg17904786 [ GTF3A ], cg10285526 [ intergenic ], cg05532093 [ NFATC1 ], cg16891376 [ FLT3LG ], and cg05352541 [ PITPNB ]. Their NIE estimates (95% CI) ranged from -0.014 (-0.037, -0.0003) for cg21682780 [ intergenic ] to -0.022 (-0.053, -0.0014) for cg05532093 [ NFATC1 ] ( Figure S6 ). When considering all nine CpG sites jointly, the estimated joint NIE was -0.150 (95% CI: -0.212, -0.095), which was similar to our main analysis. 3.5 Look-up of noteworthy CpG sites in cross-tissue databases and pathway enrichment analysis We performed a blood-brain concordance analysis for the eight CpG sites identified from the high-dimensional mediation analyses that also exhibited significant NIEs in causal mediation analysis, to evaluate how well DNAm patterns in peripheral blood reflect those in brain tissue. Using the BECon tool, four CpG sites demonstrated the highest percentile mean correlations: cg17904786 [ GTF3A ] (90%), cg16891376 [ FLT3LG ] (90%), cg10285526 [intergenic] (75-90%), and cg05532093 [NFATC1] (50-75%) ( Table S4 ). The Gene Expression Omnibus Database [Accession code GSE111165] confirmed the significant blood-brain concordance for two of these CpG sites (cg10285526 [intergenic] and cg16891376 [ FLT3LG ]) ( Table S5 ). A gene ontology (GO) analysis was conducted for the 898 CpG sites filtered from the single-mediator analysis ( Figure 1 ). After correcting for multiple testing (FDR < 0.05), no GO terms or KEGG pathways exhibited significant overrepresentation of genes containing differentially methylated CpGs. The 295 nominally significant GO terms (raw p<0.05) and the top 20 KEGG pathways were summarized in the Supplement ( Table S6, S7 ). The highest-ranking pathway identified was the arginine and proline metabolism pathway, with three differentially methylated genes ( ALDH9A1 , P4HA3 , P4HA2 ) detected by one high-dimensional mediation methods ( Table S3 ). 4 Discussion In this study, we conducted a multi-stage high-dimensional mediation analysis to investigate whether differential DNAm levels in blood mediate the previously identified association between PM 2.5 exposure and AD CSF biomarker Aβ 42 in 536 cognitively normal EHBS participants[ 9 ]. Using an analytical pipeline incorporating single-mediator analysis and multiple high-dimensional mediation methods, we identified nine noteworthy CpG sites mediating the adverse effect of PM 2.5 on Aβ 42 . Causal mediation analysis confirmed significant mediation effects for eight CpG sites, and weak correlation among the nine CpGs suggests independent mediation effects ( Figure S7 ). Six CpG sites were annotated to genes ( NFATC1 , SLC7A11 , FLT3LG , PITPNB , BEND7 , and GTF3A ) related to neurodegeneration and air pollution stress responses, suggesting their roles as biologically plausible mediators. To our knowledge, this is one of the first high-dimensional mediation studies examining DNAm as mediators between PM 2.5 exposure and AD biomarkers in a cognitively normal cohort. Among the CpGs identified, cg05532093 [ NFATC1 ] was the most significant and robust mediator ( Figures 3 , S6 ). NFATC1 encodes a transcription factor regulated by the Ca²⁺-dependent protein phosphatase calcineurin (CaN), functioning through the CaN/NFAT signaling pathway[ 73 ]. This pathway regulates pro-inflammatory genes that regulate immune response, including cytokines IL-6 , TNF-α , and the apoptotic factor FasL [ 74 ]. While cg05532093 itself has not been previously linked to AD, several CpGs within NFATC1 have been associated with AD-related CSF biomarkers, including Aβ 42 , Aβ 38 , p-tau, and neurofilament light chain, in the blood-based EWAS from the European Medical Information Framework for Alzheimer’s Disease (EMIF-AD) study[ 75 ]. Brain-based EWASs have also reported differentially methylated sites within NFATC1 in association with Braak staging [ 22 , 71 , 76 ]. In line with the reported cross-tissue association between NFATC1 and AD, our blood-brain concordance analysis showed that cg05532093 exhibited moderate correlation (50th–75th percentile) between blood and brain DNAm ( Table S4 ), suggesting that its methylation in blood may reflect similar epigenetic changes in the brain. Additionally, multiple NFATC1 CpGs have been associated with air pollution exposure in blood-based EWAS, including cg06937978 and cg18092363, which were linked to PM 2.5 [ 32 , 33 ] and NO 2 [ 77 ]. Biochemical and molecular studies also suggest CaN/NFAT signaling can be a mediator of Aβ-induced neurotoxicity in both early and advanced stages of AD progression[ 73 , 78 ], and inhibitors targeting this pathway (e.g., cyclosporine A, tacrolimus) has shown promise in alleviating AD symptoms [ 74 ]. Another significant mediator, cg00534274, is located in the first exon of SLC7A11 , encoding a subunit of the cystine–glutamate antiporter system Xc-, which is critical for antioxidant defenses. DNAm at this site has been linked to systemic inflammation via C-reactive protein (CRP) levels in EWASs[ 79 ], suggesting a link between PM 2.5 -induced oxidative stress and epigenetic regulation of SLC7A11 . This antiporter also modulates glutamate release, an essential process in excitatory neurotransmission that is disrupted in neurodegenerative diseases[ 80 ]. Other CpGs within SLC7A11 , including cg07661704 and cg06690548, have been associated with AD and Parkinson’s disease in blood-based EWASs[ 75 , 81 ]. A recent study reported increased SLC7A11 mRNA and pLG72 (D-amino acids oxidase modulator) proteins levels in plasma from AD patients, with the combination of these biomarkers showing high specificity and sensitivity for AD detection (AUC=0.882)[ 82 ]. Cg11676342 [ BEND7 ] was another noteworthy mediator. BEND7 encodes a protein involved in protein-DNA and protein-protein interactions. One CpG (cg07995537) within BEND7 was previously associated with abnormal CSF t-tau levels in a blood-based EWAS in differentiating AD, MCI patients to healthy controls[ 75 ]. Bioinformatics study have identified BEND7 as part of an AD-specific molecular network[ 83 ], though its role in AD remains poorly understood. Additionally, BEND7 CpGs have been associated with BMI[ 84 , 85 ], CRP levels[ 79 ], rheumatoid arthritis[ 86 ], but its link to PM 2.5 remains unexplored. Cg17904786 [ GTF3A ], located in the promoter region of a gene encoding General Transcription Factor IIIA, was also a significant mediator. This CpG site has been associated with age[ 87 ] and rheumatoid arthritis[ 86 ], suggesting its relevance to age-related processes. In the context of AD, another CpG in GTF3A (cg22180069) was among the 1000 differentially methylated sites associated with CSF Nfl levels in AD patients[ 75 ]. Transcriptomic studies have also identified GTF3A as differentially expressed in AD brains with varying severities, suggesting a role in AD progression[ 88 ]. Cg16891376 [ FLT3LG ], located in the promoter region of a gene encoding a cytokine involved in dendritic cell development, was also a noteworthy mediator. A recent study identified FLT3LG as one of the top-ranked CSF immune activation markers associated with neuronal injury in preclinical AD[ 89 ]. These findings suggest that changes in CSF FLT3LG and its DNAm status may occur during the presymptomatic stages of AD pathogenesis[ 89 , 90 ]. Cg05352541 [ PITPNB ], encoding a phosphatidylinositol transfer protein, plays critical roles in lipid signaling and membrane dynamics. Two CpG sites within PITPNB has been associated with tobacco smoking in blood-based EWASs[ 91 , 92 ]. While its connection to AD remains underexplored, transcriptomic analyses have identified PITPNB among the top 300 hippocampal genes potentially involved in AD neuropathology[ 93 ]. Additionally, a proteomic study using brain tissue-derived dataset reported PITPNB as a predictive biomarker distinguishing asymptomatic AD from controls[ 94 ]. Our KEGG pathway enrichment analysis identified the arginine and proline metabolism pathway as the top-ranked pathway associated with PM 2.5 exposure and CSF Aβ 42 levels. This pathway have been implicated in AD progression, with metabolomics studies detecting alterations in arginine and proline metabolism across multiple tissues, including the brain, CSF, and plasma[ 95 , 96 ]. Notably, this pathway is also a well-established sensitive marker of air pollution and traffic-related PM exposure[ 97 , 98 ], suggesting perturbation in arginine and proline metabolism as a potential link between PM 2.5 -driven oxidative stress and AD-related neurodegeneration. This study has several notable strengths. First, we employed a rigorous analytical pipeline that integrated single-mediator analysis, multiple high-dimensional mediation methods, and causal mediation analysis. Previous studies using multiple high-dimensional methods have reported low concordance across methods and limited statistical power to detect mediation effects[ 34 , 54 , 55 ]. To address these challenges, we followed recommendations from Clark-Boucher et al. (2023) to compare CpG sites results across five distinct methods (DACT, HIMA1, HIM2, HDMA, and MedFix) with nominal significance (raw p-value <0.05)[ 44 ]. We considered CpG sites as noteworthy active mediators if they were identified by at least two methods, which reduced the risk for false discoveries. To further validate our findings, we assessed the indirect effects of these CpG sites using causal mediation analysis that accounted exposure-mediator interactions[ 42 , 43 ], confirming statistically significant mediation effects for eight of them. This approach added robustness and confidence to our results. We also estimated the joint mediation effects of all nine identified CpG sites, providing an overall view of their total mediation effect. Second, the EHBS is among the largest prospective cohort studies that include CSF samples from cognitively normal individuals[ 45 ]. The comprehensive nature of our outcome assessment—a highly invasive and difficult-to-obtain biological fluid— offers a unique and valuable opportunity to explore potential associations between PM 2.5 exposure, differential DNAm and AD CSF biomarkers in cognitively normal individuals. Third, our use of a high-resolution PM 2.5 exposure model enabled precise characterization of spatial variability in individual exposure, reducing potential measurement error. However, our study has several limitations. First, our analyses were constrained by a relatively small sample size (N=536), which may have limited the statistical power to detect DNAm mediation effects on CSF Aβ 42 levels in cognitively normal EHBS participants. The limited sample size might have also reduced our ability to detect total effects between ambient PM 2.5 exposure and CSF Aβ 42 concentrations. Notably, while a negative but non-significant total effect was observed in the current analysis, a significant negative total effect was detected when the analysis was expanded to the larger EHBS participant sample not restricted to availability of DNAm data (N=1,113)[ 9 ]. Future studies with larger sample size are essential to confirm and validate these findings. Second, DNAm signatures are tissue- and cell-type specific, and DNAm in brain tissue would be more directly relevant for studying CSF AD biomarkers. Although CSF surrounds the brain, it has been reported ∼80% of CSF proteins originate from blood plasma that crosses the blood-brain barrier[ 99 ]. Thus, it is plausible that the DNAm alterations we observed in blood could mediate CSF transcriptomic and proteomic changes. Importantly, peripheral tissue samples such as blood are far more feasible to collect than brain tissue, enabling the possibility of gathering longitudinal samples to investigate temporal changes in DNAm, AD biomarkers, and cognitive decline. Third, the temporal sequence between mediators (DNAm changes) and outcome (CSF Aβ 42 concentrations) could not be clearly defined, as both biospecimen were collected at the baseline visit. However, as EHBS is an ongoing cohort, future data may allow for a better longitudinal study design to address this limitation. Finally, while our findings suggest a potential mediation role of neuroinflammatory pathways in the association PM 2.5 exposure and CSF AD biomarkers, the study did not directly measure the proinflammatory factors or the expression of related genes in blood or CSF. Future studies incorporating multi-omics approaches —including transcriptomics, proteomics, and metabolomics— are warranted to validate and expand upon our findings. 5 Conclusion Using a combination of single-mediator analysis, high-dimensional mediation analysis, and causal mediation analysis, this study identified differential DNAm in blood as a significant mediator of the association between PM 2.5 exposure and CSF Aβ 42 concentrations in the EHBS cohort of 536 cognitively normal individuals. Nine differentially methylated CpG sites were identified as noteworthy mediators, with their corresponding genes implicated in pathways related to neuroinflammation and neuroinflammation-mediated necroptosis. These findings provide valuable insights into the biological mechanisms underlying PM 2.5 toxicity during the early stages of AD pathogenesis. They also highlight the potential utility of blood DNAm changes as accessible biomarkers for detecting AD-related alterations and environmental exposures. While the relatively small sample size and the current lack of a universally accepted gold-standard approach for high-dimensional mediation analysis suggest this study serves as an initial investigation, the results are novel and promising. Further studies are warranted to replicate and expand these results, enabling a deeper understanding of how DNAm and other biological pathways mediate the impact of air pollution exposure on AD development. Data Availability All data produced in the present study are available upon reasonable request to the authors Conflict of Interest The authors declare no competing interests. Author disclosures are available in the Supporting Information. Consent Statement The Emory Healthy Brain Study was approved by the Institutional Review Board of Emory University Medical Center. All participants provided written informed consent. Acknowledgments We gratefully acknowledge the research volunteers and staff of the Goizueta Alzheimer’s Disease Research Center at Emory University, and Emory Healthy Brain Study for their participation and contributions. This work was supported by I01 BX003853 (APW), I01 BX005686 (APW), IK4 BX005219 (APW), P30 AG066511 (AIL, JJL), R01 AG056533 (APW, TSW), R01 AG070937 (JJL), R01 AG072120 (APW, TSW), R01 AG075827 (APW, TSW), R01 AG079170 (AH, TSW), U01 AG046161 (AIL), U01 AG061356 (AIL), U01 AG061357 (AIL), U01 AG088425 (AH, TSW, DL), R01AG087250 (AH, DL) References 1. ↵ Better , M.A. , Alzheimer’s disease facts and figures . Alzheimers Dement , 2023 . 19 ( 4 ): p. 1598 – 1695 . OpenUrl CrossRef PubMed 2. ↵ Lanctôt , K.L. , et al. , Burden of illness in people with Alzheimer’s disease: a systematic review of epidemiology, comorbidities and mortality . The Journal of Prevention of Alzheimer’s Disease , 2024 . 11 ( 1 ): p. 97 – 107 . OpenUrl 3. ↵ Shou , Y. , et al. , A review of the possible associations between ambient PM2. 5 exposures and the development of Alzheimer’s disease . Ecotoxicology and Environmental Safety , 2019 . 174 : p. 344 – 352 . OpenUrl CrossRef PubMed 4. Quan , M. , et al. , Genetic phenotypes of alzheimer’s disease: Mechanisms and potential therapy . Phenomics , 2023 . 3 ( 4 ): p. 333 – 349 . OpenUrl PubMed 5. ↵ Hoogmartens , J. , R. Cacace , and C. Van Broeckhoven , Insight into the genetic etiology of Alzheimer’s disease: A comprehensive review of the role of rare variants . Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring , 2021 . 13 ( 1 ): p. e12155 . OpenUrl 6. ↵ Fu , P. and K.K.L. Yung , Air pollution and Alzheimer’s disease: a systematic review and meta-analysis . Journal of Alzheimer’s Disease , 2020 . 77 ( 2 ): p. 701 – 714 . OpenUrl 7. ↵ Thiankhaw , K. , N. Chattipakorn , and S.C. Chattipakorn , PM2. 5 exposure in association with AD-related neuropathology and cognitive outcomes . Environmental Pollution , 2022 . 292 : p. 118320 . OpenUrl CrossRef PubMed 8. ↵ Alemany , S. , et al. , Associations between air pollution and biomarkers of Alzheimer’s disease in cognitively unimpaired individuals . Environment international , 2021 . 157 : p. 106864 . OpenUrl CrossRef PubMed 9. ↵ Casey , E. , et al. , Association between fine particulate matter exposure and cerebrospinal fluid biomarkers of Alzheimer’s disease among a cognitively healthy population-based cohort . Environmental Health Perspectives , 2024 . 132 ( 4 ): p. 047001 . OpenUrl PubMed 10. ↵ Li , M. , et al. , Association between air pollution and CSF sTREM2 in cognitively normal older adults: The CABLE study . Annals of Clinical and Translational Neurology , 2022 . 9 ( 11 ): p. 1752 – 1763 . OpenUrl 11. ↵ De Plano , L.M. , et al. , Epigenetic Changes in Alzheimer’s Disease: DNA Methylation and Histone Modification . Cells , 2024 . 13 ( 8 ): p. 719 . OpenUrl 12. ↵ Weymouth , L. , A.R. Smith , and K. Lunnon , DNA Methylation in Alzheimer’s Disease . 2024 . 13. ↵ Iwata , A. , et al. , Altered CpG methylation in sporadic Alzheimer’s disease is associated with APP and MAPT dysregulation . Human molecular genetics , 2014 . 23 ( 3 ): p. 648 – 656 . OpenUrl CrossRef PubMed 14. Barrachina , M. and I. Ferrer , DNA methylation of Alzheimer disease and tauopathy-related genes in postmortem brain . Journal of Neuropathology & Experimental Neurology , 2009 . 68 ( 8 ): p. 880 – 891 . OpenUrl PubMed 15. ↵ Wang , S.-C. , B. Oelze , and A. Schumacher , Age-specific epigenetic drift in late-onset Alzheimer’s disease . PloS one , 2008 . 3 ( 7 ): p. e2698 . OpenUrl CrossRef PubMed 16. ↵ Yu , L. , et al. , Association of Brain DNA methylation in SORL1, ABCA7, HLA-DRB5, SLC24A4, and BIN1 with pathological diagnosis of Alzheimer disease . JAMA neurology , 2015 . 72 ( 1 ): p. 15 – 24 . OpenUrl PubMed 17. ↵ Bezuch , N. , et al. , Superior frontal gyrus TOMM40-APOE locus DNA methylation in Alzheimer’s disease . Journal of Alzheimer’s Disease Reports , 2021 . 5 ( 1 ): p. 275 – 282 . OpenUrl 18. ↵ Bakulski , K.M. , et al. , Genome-wide DNA methylation differences between late-onset Alzheimer’s disease and cognitively normal controls in human frontal cortex . Journal of Alzheimer’s Disease , 2012 . 29 ( 3 ): p. 571 – 588 . OpenUrl 19. ↵ Sanchez−Mut , J.V. , et al. , Promoter hypermethylation of the phosphatase DUSP22 mediates PKA−dependent TAU phosphorylation and CREB activation in Alzheimer’s disease . Hippocampus , 2014 . 24 ( 4 ): p. 363 – 368 . OpenUrl CrossRef PubMed 20. ↵ De Jager , P.L. , et al. , Alzheimer’s disease: early alterations in brain DNA methylation at ANK1, BIN1, RHBDF2 and other loci . Nature neuroscience , 2014 . 17 ( 9 ): p. 1156 – 1163 . OpenUrl CrossRef PubMed 21. ↵ Lunnon , K. , et al. , Methylomic profiling implicates cortical deregulation of ANK1 in Alzheimer’s disease . Nature neuroscience , 2014 . 17 ( 9 ): p. 1164 – 1170 . OpenUrl CrossRef PubMed 22. ↵ Smith , R.G. , et al. , Elevated DNA methylation across a 48−kb region spanning the HOXA gene cluster is associated with Alzheimer’s disease neuropathology . Alzheimer’s & Dementia , 2018 . 14 ( 12 ): p. 1580 – 1588 . OpenUrl 23. ↵ Li , Q.S. , Y. Sun , and T. Wang , Epigenome-wide association study of Alzheimer’s disease replicates 22 differentially methylated positions and 30 differentially methylated regions . Clinical epigenetics , 2020 . 12 : p. 1 – 14 . OpenUrl CrossRef 24. ↵ Zhang , L. , et al. , Epigenome-wide meta-analysis of DNA methylation differences in prefrontal cortex implicates the immune processes in Alzheimer’s disease . Nature communications , 2020 . 11 ( 1 ): p. 6114 . OpenUrl PubMed 25. ↵ Smith , R.G. , et al. , A meta-analysis of epigenome-wide association studies in Alzheimer’s disease highlights novel differentially methylated loci across cortex . Nature communications , 2021 . 12 ( 1 ): p. 3517 . OpenUrl PubMed 26. ↵ De Prins , S. , et al. , Influence of ambient air pollution on global DNA methylation in healthy adults: a seasonal follow-up . Environment international , 2013 . 59 : p. 418 – 424 . OpenUrl CrossRef PubMed Web of Science 27. ↵ Bellavia , A. , et al. , DNA hypomethylation, ambient particulate matter, and increased blood pressure: findings from controlled human exposure experiments . Journal of the American Heart Association , 2013 . 2 ( 3 ): p. e000212 . OpenUrl Abstract / FREE Full Text 28. ↵ Wu , Y. , et al. , Air pollution and DNA methylation in adults: a systematic review and meta-analysis of observational studies . Environmental pollution , 2021 . 284 : p. 117152 . OpenUrl CrossRef PubMed 29. ↵ Bind , M.-A. , et al. , Air pollution and gene-specific methylation in the Normative Aging Study: association, effect modification, and mediation analysis . Epigenetics , 2014 . 9 ( 3 ): p. 448 – 458 . OpenUrl CrossRef PubMed 30. ↵ Li , R. , et al. , PM2. 5-induced DNA oxidative stress in A549 cells and regulating mechanisms by GST DNA methylation and Keap1/Nrf2 pathway . Toxicology Mechanisms and Methods , 2024 . 34 ( 5 ): p. 517 – 526 . OpenUrl 31. ↵ Wang , C. , et al. , Long-term ambient fine particulate matter and DNA methylation in inflammation pathways: results from the Sister Study . Epigenetics , 2020 . 15 ( 5 ): p. 524 – 535 . OpenUrl CrossRef PubMed 32. ↵ Panni , T. , et al. , Genome-wide analysis of DNA methylation and fine particulate matter air pollution in three study populations: KORA F3, KORA F4, and the normative aging study . Environmental health perspectives , 2016 . 124 ( 7 ): p. 983 – 990 . OpenUrl CrossRef PubMed 33. ↵ Eze , I.C. , et al. , Genome-wide DNA methylation in peripheral blood and long-term exposure to source-specific transportation noise and air pollution: the SAPALDIA study . Environmental health perspectives , 2020 . 128 ( 6 ): p. 067003 . OpenUrl CrossRef PubMed 34. ↵ Li , Z. , et al. , Differential DNA methylation in the brain as potential mediator of the association between traffic-related PM2. 5 and neuropathology markers of Alzheimer’s disease . Alzheimer’s & Dementia , 2024 . 20 ( 4 ): p. 2538 – 2551 . OpenUrl 35. ↵ Rani , P. , S. Krishnan , and C. Rani Cathrine , Study on analysis of peripheral biomarkers for Alzheimer’s disease diagnosis . Frontiers in neurology , 2017 . 8 : p. 328 . OpenUrl PubMed 36. ↵ Baron , R.M. and D.A. Kenny , The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations . Journal of personality and social psychology , 1986 . 51 ( 6 ): p. 1173 . OpenUrl CrossRef PubMed Web of Science 37. ↵ Liu , Z. , et al. , Large-scale hypothesis testing for causal mediation effects with applications in genome-wide epigenetic studies . Journal of the American Statistical Association , 2022 . 117 ( 537 ): p. 67 – 81 . OpenUrl PubMed 38. ↵ Zhang , H. , et al. , Estimating and testing high-dimensional mediation effects in epigenetic studies . Bioinformatics , 2016 . 32 ( 20 ): p. 3150 – 3154 . OpenUrl CrossRef PubMed 39. ↵ Perera , C. , et al. , HIMA2: high-dimensional mediation analysis and its application in epigenome-wide DNA methylation data . BMC bioinformatics , 2022 . 23 ( 1 ): p. 296 . OpenUrl CrossRef PubMed 40. ↵ Gao , Y. , et al. , Testing mediation effects in high-dimensional epigenetic studies . Frontiers in genetics , 2019 . 10 : p. 1195 . OpenUrl PubMed 41. ↵ Zhang , Q ., High-dimensional mediation analysis with applications to causal gene identification . Statistics in biosciences , 2022 . 14 ( 3 ): p. 432 – 451 . OpenUrl CrossRef 42. ↵ Shi , B. , et al. , CMAverse: a suite of functions for reproducible causal mediation analyses . Epidemiology , 2021 . 32 ( 5 ): p. e20 – e22 . OpenUrl CrossRef PubMed 43. ↵ VanderWeele , T. and S. Vansteelandt , Mediation analysis with multiple mediators . Epidemiologic methods , 2014 . 2 ( 1 ): p. 95 – 115 . OpenUrl PubMed 44. ↵ Clark-Boucher , D. , et al. , Methods for mediation analysis with high-dimensional DNA methylation data: Possible choices and comparisons . PLoS genetics , 2023 . 19 ( 11 ): p. e1011022 . OpenUrl 45. ↵ Goetz , M.E. , et al. , Rationale and design of the emory healthy aging and emory healthy brain studies . Neuroepidemiology , 2019 . 53 ( 3-4 ): p. 187 – 200 . OpenUrl CrossRef PubMed 46. ↵ Huels , A. , et al. , Epigenome-wide association study of cerebrospinal fluid-based biomarkers of Alzheimer’s disease in cognitively normal individuals . medRxiv , 2025 : p. 2025.02. 04.25321657. 47. ↵ Aryee , M.J. , et al. , Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays . Bioinformatics , 2014 . 30 ( 10 ): p. 1363 – 1369 . OpenUrl CrossRef PubMed Web of Science 48. ↵ Salas , L.A. , et al. , An optimized library for reference-based deconvolution of whole-blood biospecimens assayed using the Illumina HumanMethylationEPIC BeadArray . Genome biology , 2018 . 19 : p. 1 – 14 . OpenUrl CrossRef PubMed 49. ↵ Di , Q. , et al. , Daily and annual PM2. 5 concentrations for the contiguous United States, 1-km grids, v1 (2000-2016) . NASA Socioeconomic Data and Applications Center (SEDAC) , 2021 . 50. ↵ Di , Q. , et al. , An ensemble-based model of PM2. 5 concentration across the contiguous United States with high spatiotemporal resolution . Environment international , 2019 . 130 : p. 104909 . OpenUrl CrossRef PubMed 51. ↵ Singh , G.K ., Area deprivation and widening inequalities in US mortality, 1969–1998 . American journal of public health , 2003 . 93 ( 7 ): p. 1137 – 1143 . OpenUrl CrossRef PubMed Web of Science 52. ↵ Li , Z. , et al. , Neighborhood characteristics as confounders and effect modifiers for the association between air pollution exposure and subjective cognitive functioning . Environmental research , 2022 . 212 : p. 113221 . OpenUrl 53. ↵ Zeng , P. , Z. Shao , and X. Zhou , Statistical methods for mediation analysis in the era of high-throughput genomics: current successes and future challenges . Computational and structural biotechnology journal , 2021 . 19 : p. 3209 – 3224 . OpenUrl 54. ↵ Feil , D. , et al. , DNA methylation as a potential mediator of the association between indoor air pollution and neurodevelopmental delay in a South African birth cohort . Clinical Epigenetics , 2023 . 15 ( 1 ): p. 31 . OpenUrl PubMed 55. ↵ Abrishamcar , S. , et al. , DNA methylation as a potential mediator of the association between prenatal tobacco and alcohol exposure and child neurodevelopment in a South African birth cohort . Translational psychiatry , 2022 . 12 ( 1 ): p. 418 . OpenUrl PubMed 56. ↵ Robins , J.M. and S. Greenland , Identifiability and exchangeability for direct and indirect effects . Epidemiology , 1992 . 3 ( 2 ): p. 143 – 155 . OpenUrl CrossRef PubMed Web of Science 57. ↵ VanderWeele , T.J. and S. Vansteelandt , Conceptual issues concerning mediation, interventions and composition . Statistics and its Interface , 2009 . 2 ( 4 ): p. 457 – 468 . OpenUrl CrossRef Web of Science 58. ↵ Valeri , L. and T.J. VanderWeele , Mediation analysis allowing for exposure–mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros . Psychological methods , 2013 . 18 ( 2 ): p. 137 . OpenUrl CrossRef PubMed Web of Science 59. ↵ Fan , J. and J. Lv , Sure independence screening for ultrahigh dimensional feature space . Journal of the Royal Statistical Society Series B: Statistical Methodology , 2008 . 70 ( 5 ): p. 849 – 911 . OpenUrl CrossRef PubMed 60. ↵ Zhang , C.-H ., Nearly unbiased variable selection under minimax concave penalty . The Annals of Statistics , 2010 . 38 ( 2 ): p. 894 . OpenUrl 61. ↵ Berger , R.L. and J.C. Hsu , Bioequivalence trials, intersection-union tests and equivalence confidence sets . Statistical Science , 1996 . 11 ( 4 ): p. 283 – 319 . OpenUrl CrossRef Web of Science 62. ↵ Huang , Y.-T ., JOINT SIGNIFICANCE TESTS FOR MEDIATION EFFECTS OF SOCIOECONOMIC ADVERSITY ON ADIPOSITY VIA EPIGENETICS . The Annals of Applied Statistics , 2018 . 12 ( 3 ): p. 1535 – 1557 . OpenUrl 63. ↵ Zhang , C.-H. and S.S. Zhang , Confidence intervals for low dimensional parameters in high dimensional linear models . Journal of the Royal Statistical Society Series B: Statistical Methodology , 2014 . 76 ( 1 ): p. 217 – 242 . OpenUrl 64. ↵ Dai , J.Y. , J.L. Stanford , and M. LeBlanc , A multiple-testing procedure for high-dimensional mediation hypotheses . Journal of the American Statistical Association , 2022 . 117 ( 537 ): p. 198 – 213 . OpenUrl CrossRef PubMed 65. ↵ Zou , H ., The adaptive lasso and its oracle properties . Journal of the American statistical association , 2006 . 101 ( 476 ): p. 1418 – 1429 . OpenUrl CrossRef Web of Science 66. ↵ VanderWeele , T.J. , S. Vansteelandt , and J.M. Robins , Effect decomposition in the presence of an exposure-induced mediator-outcome confounder . Epidemiology , 2014 . 25 ( 2 ): p. 300 – 306 . OpenUrl CrossRef PubMed Web of Science 67. ↵ Gleason , C.E. , et al. , Alzheimer’s disease biomarkers in Black and non−Hispanic White cohorts: a contextualized review of the evidence . Alzheimer’s & Dementia , 2022 . 18 ( 8 ): p. 1545 – 1564 . OpenUrl 68. ↵ Edgar , R. , et al. , BECon: a tool for interpreting DNA methylation findings from blood in the context of brain . Translational Psychiatry , 2017 . 7 ( 8 ): p. e1187 – e1187 . OpenUrl PubMed 69. ↵ Phipson , B. , J. Maksimovic , and A. Oshlack , missMethyl: an R package for analyzing data from Illumina’s HumanMethylation450 platform . Bioinformatics , 2015 . 32 ( 2 ): p. 286 – 288 . OpenUrl PubMed 70. ↵ Hansen , K. and M. Aryee , IlluminaHumanMethylationEPICmanifest: Manifest for Illumina’s EPIC methylation arrays . R package version , 2016 . 3 ( 0 ). 71. ↵ Battram , T. , et al. , The EWAS Catalog: a database of epigenome-wide association studies . Wellcome open research , 2022 . 7 . 72. ↵ EPA . National Ambient Air Quality Standards (NAAQS) for PM . 2024 [cited 2024 September 20]; Available from: https://www.epa.gov/pm-pollution/national-ambient-air-quality-standards-naaqs-pm#rule-summary . 73. ↵ Abdul , H.M. , et al. , NFATs and Alzheimer’s disease . Molecular and Cellular Pharmacology , 2010 . 2 ( 1 ): p. 7 . OpenUrl 74. ↵ Mackiewicz , J. , M. Lisek , and T. Boczek , Targeting CaN/NFAT in Alzheimer’s brain degeneration . Frontiers in Immunology , 2023 . 14 : p. 1281882 . OpenUrl PubMed 75. ↵ Smith , R.G. , et al. , Blood DNA methylomic signatures associated with CSF biomarkers of Alzheimer’s disease in the EMIF−AD study . Alzheimer’s & Dementia , 2024 . 20 ( 10 ): p. 6722 – 6739 . OpenUrl 76. ↵ Gasparoni , G. , et al. , DNA methylation analysis on purified neurons and glia dissects age and Alzheimer’s disease-specific changes in the human cortex . Epigenetics & chromatin , 2018 . 11 : p. 1 – 19 . OpenUrl CrossRef PubMed 77. ↵ de FC Lichtenfels , A.J. , et al. , Long-term air pollution exposure, genome-wide DNA methylation and lung function in the LifeLines cohort study . Environmental health perspectives , 2018 . 126 ( 2 ): p. 027004 . OpenUrl CrossRef PubMed 78. ↵ Abdul , H.M. , et al. , Cognitive decline in Alzheimer’s disease is associated with selective changes in calcineurin/NFAT signaling . Journal of Neuroscience , 2009 . 29 ( 41 ): p. 12957 – 12969 . OpenUrl Abstract / FREE Full Text 79. ↵ Hillary , R.F. , et al. , Blood-based epigenome-wide analyses of chronic low-grade inflammation across diverse population cohorts . Cell Genomics , 2024 . 4 ( 5 ). 80. ↵ Moustapha , A ., Neurodegenerative diseases: potential effect of glutathione, in Glutathione System and Oxidative Stress in Health and Disease . 2020 , IntechOpen. 81. ↵ Vallerga , C.L. , et al. , Analysis of DNA methylation associates the cystine–glutamate antiporter SLC7A11 with risk of Parkinson’s disease . Nature communications , 2020 . 11 ( 1 ): p. 1238 . OpenUrl PubMed 82. ↵ Lane , H.-Y. and C.-H. Lin , Diagnosing Alzheimer’s disease specifically and sensitively with pLG72 and cystine/glutamate antiporter SLC7A11 as blood biomarkers . International Journal of Neuropsychopharmacology , 2023 . 26 ( 1 ): p. 1 – 8 . OpenUrl PubMed 83. ↵ Hu , Y.-S. , et al. , Analyzing the genes related to Alzheimer’s disease via a network and pathway-based approach . Alzheimer’s research & therapy , 2017 . 9 : p. 1 – 15 . OpenUrl CrossRef 84. ↵ Sun , D. , et al. , Body mass index drives changes in DNA methylation: a longitudinal study . Circulation research , 2019 . 125 ( 9 ): p. 824 – 833 . OpenUrl CrossRef PubMed 85. ↵ Geurts , Y. , et al. , Novel associations between blood DNA methylation and body mass index in middle-aged and older adults . International journal of obesity , 2018 . 42 ( 4 ): p. 887 – 896 . OpenUrl 86. ↵ Liu , Y. , et al. , Epigenome-wide association data implicate DNA methylation as an intermediary of genetic risk in rheumatoid arthritis . Nature biotechnology , 2013 . 31 ( 2 ): p. 142 – 147 . OpenUrl CrossRef PubMed 87. ↵ Mulder , R.H. , et al. , Epigenome-wide change and variation in DNA methylation in childhood: trajectories from birth to late adolescence . Human molecular genetics , 2021 . 30 ( 1 ): p. 119 – 134 . OpenUrl CrossRef PubMed 88. ↵ Kong , W. , et al. , Differences of immune disorders between Alzheimer’s disease and breast cancer based on transcriptional regulation . PloS one , 2017 . 12 ( 7 ): p. e0180337 . OpenUrl PubMed 89. ↵ Shue , F. , et al. , CSF biomarkers of immune activation and Alzheimer’s disease for predicting cognitive impairment risk in the elderly . Science Advances , 2024 . 10 ( 14 ): p. eadk3674 . OpenUrl PubMed 90. ↵ Del Campo , M. , et al. , CSF proteome profiling across the Alzheimer’s disease spectrum reflects the multifactorial nature of the disease and identifies specific biomarker panels . Nature aging , 2022 . 2 ( 11 ): p. 1040 – 1053 . OpenUrl PubMed 91. ↵ Zeilinger , S. , et al. , Tobacco smoking leads to extensive genome-wide changes in DNA methylation . PloS one , 2013 . 8 ( 5 ): p. e63812 . OpenUrl CrossRef PubMed 92. ↵ Joehanes , R. , et al. , Epigenetic signatures of cigarette smoking . Circulation: cardiovascular genetics , 2016 . 9 ( 5 ): p. 436 – 447 . OpenUrl Abstract / FREE Full Text 93. ↵ Zhang , L. , et al. , Potential hippocampal genes and pathways involved in Alzheimer’s disease: a bioinformatic analysis . Genet Mol Res , 2015 . 14 ( 2 ): p. 7218 – 7232 . OpenUrl CrossRef PubMed 94. ↵ Tandon , R. , et al. , Machine learning selection of most predictive brain proteins suggests role of sugar metabolism in Alzheimer’s disease . Journal of Alzheimer’s Disease , 2023 . 92 ( 2 ): p. 411 – 424 . OpenUrl 95. ↵ Ambeskovic , M. , et al. , Metabolomic signatures of Alzheimer’s disease indicate brain region-specific neurodegenerative progression . International journal of molecular sciences , 2023 . 24 ( 19 ): p. 14769 . OpenUrl PubMed 96. ↵ Xie , K. , et al. , High-throughput metabolomics for discovering potential biomarkers and identifying metabolic mechanisms in aging and Alzheimer’s disease . Frontiers in cell and developmental biology , 2021 . 9 : p. 602887 . OpenUrl 97. ↵ Liang , D. , et al. , Perturbations of the arginine metabolome following exposures to traffic-related air pollution in a panel of commuters with and without asthma . Environment international , 2019 . 127 : p. 503 – 513 . OpenUrl CrossRef PubMed 98. ↵ Liang , D. , et al. , A state-of-the-science review on high-resolution metabolomics application in air pollution health research: current progress, analytical challenges, and recommendations for future direction . Environmental Health Perspectives , 2023 . 131 ( 5 ): p. 056002 . OpenUrl PubMed 99. ↵ Schilde , L.M. , et al. , Protein variability in cerebrospinal fluid and its possible implications for neurological protein biomarker research . PLoS One , 2018 . 13 ( 11 ): p. e0206478 . OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted April 16, 2025. Download PDF Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Differential DNA methylation in blood as potential mediator of the association between ambient PM2.5 and cerebrospinal fluid biomarkers of Alzheimer’s disease among a cognitively normal population-based cohort Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Differential DNA methylation in blood as potential mediator of the association between ambient PM 2.5 and cerebrospinal fluid biomarkers of Alzheimer’s disease among a cognitively normal population-based cohort Tszshan Ma , Jiaqi Liu , Donghai Liang , Stefanie Ebelt , Kyle Steenland , Allan I. Levey , James J. Lah , Aliza P. Wingo , Thomas S. Wingo , Anke Hüls medRxiv 2025.04.15.25325910; doi: https://doi.org/10.1101/2025.04.15.25325910 Share This Article: Copy Citation Tools Differential DNA methylation in blood as potential mediator of the association between ambient PM 2.5 and cerebrospinal fluid biomarkers of Alzheimer’s disease among a cognitively normal population-based cohort Tszshan Ma , Jiaqi Liu , Donghai Liang , Stefanie Ebelt , Kyle Steenland , Allan I. Levey , James J. Lah , Aliza P. Wingo , Thomas S. Wingo , Anke Hüls medRxiv 2025.04.15.25325910; doi: https://doi.org/10.1101/2025.04.15.25325910 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Epidemiology Subject Areas All Articles Addiction Medicine (566) Allergy and Immunology (863) Anesthesia (295) Cardiovascular Medicine (4408) Dentistry and Oral Medicine (443) Dermatology (379) Emergency Medicine (606) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1505) Epidemiology (15202) Forensic Medicine (30) Gastroenterology (1119) Genetic and Genomic Medicine (6568) Geriatric Medicine (666) Health Economics (994) Health Informatics (4508) Health Policy (1365) Health Systems and Quality Improvement (1607) Hematology (537) HIV/AIDS (1262) Infectious Diseases (except HIV/AIDS) (15902) Intensive Care and Critical Care Medicine (1103) Medical Education (619) Medical Ethics (144) Nephrology (665) Neurology (6572) Nursing (345) Nutrition (998) Obstetrics and Gynecology (1139) Occupational and Environmental Health (954) Oncology (3319) Ophthalmology (967) Orthopedics (369) Otolaryngology (420) Pain Medicine (435) Palliative Medicine (129) Pathology (662) Pediatrics (1686) Pharmacology and Therapeutics (691) Primary Care Research (710) Psychiatry and Clinical Psychology (5420) Public and Global Health (9202) Radiology and Imaging (2190) Rehabilitation Medicine and Physical Therapy (1367) Respiratory Medicine (1191) Rheumatology (593) Sexual and Reproductive Health (709) Sports Medicine (529) Surgery (709) Toxicology (99) Transplantation (288) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9fe698865d65f047',t:'MTc3OTIzMDM0Nw=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();