Content uploaded by Gaetano Zazzaro
Author content
All content in this area was uploaded by Gaetano Zazzaro on Jul 19, 2024
Content may be subject to copyright.
Prominent role of PM10 but not of circulating inflammation in the link between
air pollution and the risk of neurodegenerative disorders
Alessandro Gialluisi1,2#, Simona Costanzo2, Giovanni Veronesi1, Assuntina Cembalo3,
Alfonsina Tirozzi2, Stefania Falciglia4, Moreno Ricci4, Francesco Martone3, Gaetano
Zazzaro3, Marco Mario Ferrario1, Francesco Gianfagna1,5, Chiara Cerletti2, Maria Benedetta
Donati2, Stefania Massari6, Giovanni de Gaetano2, Licia Iacoviello1,2, on behalf of the Moli-
sani Study Investigators7
1 EPIMED Research Center, Department of Medicine and Surgery, University of Insubria,
Varese, Italy
2 Department of Epidemiology and Prevention, IRCCS NEUROMED, Pozzilli, Italy
3 CIRA-Italian Aerospace Research Centre, Capua, Italy
4 UOC Governance del Farmaco, Azienda Sanitaria Regionale del Molise –ASREM, Campobasso,
Italy
5 Mediterranea Cardiocentro, Napoli, Italy
6 Department of Occupational and Environmental Medicine, Epidemiology and Hygiene,
Italian Workers' Compensation Authority (INAIL), Rome, Italy
7A complete list of the Moli-sani Study Investigators is available in Supplementary Materials
Corresponding Author:
Alessandro Gialluisi, PhD
Department of Medicine and Surgery, University of Insubria, Via Monte Generoso 71, 21100,
Varese, Italy
Department of Epidemiology and Prevention, IRCCS Neuromed, Via dell´Elettronica,
86077, Pozzilli, Italy
Email: alessandro.gialluisi@gmail.com
Conflict of interest
Nothing to declare.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.
Abstract
Background
Several studies revealed an implication of air pollution in neurodegenerative disorders,
although this link and the potential underlying mechanisms remain unclear.
Objectives
To analyze the impact of air pollution on neurodegenerative risk by testing multiple
pollutants simultaneously, along with other potential risk/protective factors, and the role of
circulating inflammation.
Methods
In the Moli-sani cohort (N=24,325; ≥35 years; 51.9% women, baseline 2005-2010), we
estimated yearly levels of exposure to nitrogen oxides, ozone, particulate matter (PM10),
sulfur dioxide and BTX hydrocarbons in 2006-2018, applying residence geo-localization of
participants and Kriging interpolation algorithm to land measurements of air pollutants.
We performed a principal component (PC) analysis of pollutant levels and tested
associations of the resulting PC scores with the incident risk of dementia (AD) and
Parkinson’s disease/parkinsonism (PD), through multivariable Cox PH regressions adjusted
for age, sex, education level, and several professional and lifestyle exposures. Moreover, we
tested whether a composite biomarker of circulating inflammation (INFLA-score) may
explain part of these associations.
Results
Over 24,308 subjects with pollution data available (51.9% women, mean age 55.8(12.0)
years), we extracted three PCs explaining ≥5% of pollution exposure variance: PC1 (38.2%,
tagging PM10), PC2 (19.5%, O3/CO/SO2), PC3 (8.5%, NOx/BTX hydrocarbons). Over a
median (IQR) follow-up of 11.2(2.0) years, we observed statistically significant associations
of PC1 with an increased risk of both AD (HR[CI] = 1.06[1.04-1.08]; 218 cases) and PD
(1.05[1.03-1.06]; 405 incident cases), independent on other covariates. These associations
were confirmed testing average PM10 levels during follow-up time (25[19-31]% and 19[15-
24]% increase of AD and PD risk, per 1 μg/m3 of PM10). INFLA-score explained a negligible
(<1%) proportion of these associations.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Discussion Air pollution – especially PM10 – is associated with increased
neurodegenerative risk in the Italian population, independent on concurring risk factors,
suggesting its reduction as a potential public health target.
Keywords: air pollution, PM10, dementia, Parkinson’s disease and parkinsonisms,
Alzheimer’s disease, circulating inflammation, C-reactive protein, granulocytes, white
blood cells
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Introduction
In the last decades, pollution – prominently air pollution – has represented a hotspot of
investigation and received increasing interest from policy makers, in view of the notable
burden that it implies for public health and welfare systems [1]. Air pollution is defined as
a combination of various compounds originating from different anthropogenic and biogenic
emission sources, ranging from particulate matter (PM, with different components based on
their aerodynamic diameter) to gases (carbon monoxide, nitrogen dioxide, etc.), and organic
chemical compounds (e.g., hydrocarbons) [2]. Not only chronic exposure to air pollution is
associated with increased cardiovascular and respiratory diseases risk [3], but recent
findings have supported a link between air pollution and neurological disorders, including
neurodegenerative disorders like dementia of different origins (hereafter called AD), and
Parkinson’s disease or parkinsonisms (hereafter called PD) [5,6]. Dementia is the most
prevalent neurodegenerative disorder worldwide, is characterized by memory loss and
cognitive impairment and may be of different types, the most frequent being vascular
dementia and Alzheimer’s disease. The latter represents 60-80% of all dementia forms [7]
and is characterized by typical accumulations of amyloid beta and tau proteins aggregates
in the brain, leading to neuronal loss in cortical areas and brain atrophy. Parkinson’s disease
represents instead the second most common neurodegenerative disease, characterized by
alpha-synuclein aggregates accumulation in dopaminergic neurons of the substantia nigra,
which leads to progressive, irreversible loss of these neurons. This manifests through
muscular rigidity, bradykinesia, tremor of resting limbs, gait and balance impairments, and
often to a progressive loss of cognitive functions [8]. Moreover, Parkinson’s disease often
shows several common characteristics with Parkinsonisms - clinical syndromes with similar
manifestations, although of different origins, like vascular parkinsonism and multiple
system atrophy [9,10]. Interestingly, both AD and PD are often comorbid, show a partial
overlap of signs and symptoms and are triggered at least in part by the same molecular
mechanisms, like circulating and neuro-inflammation [11,12].
While recent reviews and meta-analyses have reported growing evidence suggesting that
chronic exposure to air pollution – especially to particulate matter with aerodynamic
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
diameter < 2.5 μm (PM2.5) and nitric dioxide (NO2) – is associated with an increased risk of
both incident and prevalent neurodegenerative disorders like AD and PD [13,14,15,16],
these associations are often not replicated and only partly concordant across the different
studies, for several reasons. Among them, the heterogeneity of study settings probably
represents one of the main hindrances to results concordance, with very scarce longitudinal
studies carried out in large real cohorts (e.g. [17]) and many administrative cohort (e.g.
[18,19,20]), case-control studies (e.g. [21,22]) and geostatistical analyses (e.g. [23]) being
reported. Indeed, most of these studies tested a handful of pollutants simultaneously, but
only few analyzed them together with other potential factors influencing neurodegenerative
risk. As a consequence, these studies often suffer from residual confounding bias due to
scarce adjustment for other potential confounders of the association between air pollution
and neurodegenerative risk, like lifestyles and professional covariates. Other studies are
mostly based on a case-control approach – due to the relatively low prevalence of these
disorders – and hence may suffer from reverse causality bias. Overall, very few studies are
based on deeply characterized longitudinal cohorts (e.g. [17]) which may allow to
disentangle the cluster of risk/protective factors for neurodegenerative disorders, including
sociodemographic, lifestyles and professional factors.
The main aim of this work was to test how air pollution may influence neurodegenerative
risk – in particular the incident risk of the most prevalent neurodegenerative disorders like
dementia and Parkinson’s disease/parkinsonisms – independent on other potential
risk/protective factors. We did this in a general Italian population cohort, with a deep
phenotypic assessment, by simultaneously analyzing the influence of several air pollutants,
sociodemographic, lifestyle and professional exposures, over a 12 years follow-up. Thanks
to the availability of molecular blood markers, we also tested a potential role of circulating
inflammation in the links identified, looking for further support to previous evidence
reported in animal models [6,24,25].
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Subjects and Methods
Population of study
The study population consisted of participants to the Moli-sani project (N = 24,325; 51.9%
women), a cohort of Italian residents recruited from the general population of Molise Region
(Central-Southern Italy), between March 2005 and April 2010. Exclusion criteria were
pregnancy at the time of recruitment, inability to understand terms of participation, current
poly-traumas (i.e., simultaneous injury to several organs or body systems), coma, or refusal
to sign the informed consent [26]. The Moli-sani Study was approved by the ethical
committee of the Catholic University of Rome (approval nr: P99, A-931/03-138-04/CE/2004,
11 February 2004) and all the participants provided written informed consent.
Geolocalization
For this study, participants were geolocalized based on their residence address – available
for all the participants - with the use of complementary dedicated software APIs, like
Geokettle, QGIS and Here (see URLs). The data were cleaned up manually correcting the
errors detected, such as incorrect postal codes. To facilitate the geocoding process, the
addresses were transformed by replacing the abbreviations. Using an automated procedure
based on the Here API, it was possible to assign latitude and longitude coordinates for all
the subjects in the database. A handful of subjects with residence outside the Molise region
(n=17) were geolocated out of the Molise region and hence removed from the analysis. The
results of the geocoding were verified using the shapefiles of the boundaries of the
municipalities of the region. Addresses placed outside the limits of the city of residence
were manually geolocated using Google Maps. With this procedure, it was possible to
assign latitude and longitude coordinates for 24,308 subjects in the database with a high
level of confidence. The subjects were then linked to air pollution maps (built as described
below), which allowed us to estimate the amount of exposure to each pollutant in the
physical coordinates where subjects reported their residence.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Air pollution exposure
We estimated yearly levels of exposure to ten different pollutants, including Nitrogen
oxides (NOX, NO, NO2), ozone (O3), particulate matter with aerodynamic diameter < 10
μm (PM10), Sulfur dioxide (SO2), Carbon monoxide (CO) and BTX hydrocarbons (benzene,
toluene and xylene) in 2006-2018. To this end, we applied ordinary Kriging interpolation
algorithm to land measurements of air pollutants made publicly available from the regional
environmental authority (Agenzia Regionale per la Protezione Ambientale del Molise -
ARPA Molise, see URLs). Indeed, Kriging algorithm allows geostatistical data interpolation
based on available land measures, to infer also unsampled points across the spatial field [27]
(see URLs). To apply geostatistical algorithms and display analysis results on maps, we used
ESRI ArcGIS tool, a proprietary geographic information system (GIS; see URLs) which
allowed us to display air monitoring stations on the regional Molise map and finally build
maps of exposure to different air pollutants (see Figure 1). This algorithm was applied to
each pollutant for each year of the follow-up time considered, over participants with
physical coordinates available within the Molise region. This returned 24,308 subjects with
environmental exposures (each with ten pollutants levels × 13 years). Since Kriging returned
exposure intervals which were largely overlapping but with variable limits across the years,
to reduce collinearity among variables, we decided to apply a principal component analysis
(PCA), so to derive latent variables which could capture most of the shared variance across
the 130 variables.
To this end, we removed toluene levels for year 2017 – which presented a single level of
exposure. Then, we computed the mean of the two limits for each level of exposure (or rank)
and each pollutant, so to have point estimates of exposure for applying the PCA. We carried
out a preliminary Kaiser-Meyer-Olkin (KMO) test of sampling adequacy, which revealed a
very good factorability (MSA = 0.97), and a Bartlett's Test of Sphericity, which suggested a
significant discrepancy with the identity matrix, hence notable reciprocal correlations across
the variables, which further supported our choice. PCA was performed through Singular
Value Decomposition with orthogonal (varimax) rotation (built-in prcomp() function in
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
R[28]). The resulting PC scores were then tested as main exposures in survival analyses, as
detailed below.
Outcome: incident neurodegenerative risk
Definition of incident neurodegenerative disorders was carried out through linkage with
Electronic Health Records (EHRs) databases like the Italian National mortality (ReNCaM)
registry, the Molise regional registry of hospital discharge records (HDRs) and the regional
drug prescription registry, using fiscal code of each participant as unique identifier.
Specifically, Alzheimer’s Disease/Dementia (AD) and Parkinson's Disease/Parkinsonisms
(PD) were defined as the occurrence of at least one of the following events up to December
31st 2018 (Table S1): i) Parkinson’s and Alzheimer’s disease reported as main cause of death
(ICD-9 cm codes 332 and 331) in the ReNCaM registry; ii) Parkinson’s disease or any
parkinsonism, Alzheimer’s disease or other dementias reported as main cause of
hospitalization in HDRs (ICD-9 cm codes: 332, 332.0 and 332.1 for PD and 331, 331.X for
AD); iii) specific drug prescriptions for the treatment of these disorders in the regional drug
prescription registry, like antiparkinsonian (ATC code: N04) and antidementia drugs
(N06D).
Statistical analysis
All statistical analyses of the present manuscript were carried out in R (see URLs)[28]. After
removing prevalent AD (n=16) and PD (n=52) cases, as well as those participants with
missing information on incident events (0 for AD and 236 for PD), we tested incident AD/PD
risk vs air pollution exposure through multivariable Cox Proportional Hazards (PH)
regression models, implemented through the coxph() and the surv() functions of the survival
package [29]. To test potential influences of air pollution independent on other clusters of
risk/protective factors, multivariable models were built, incrementally adjusted for i) age,
sex and education level completed (baseline, Model 1); ii) professional factors like exposure
to toxic compounds and professional working class (Model 2); and iii) lifestyles (smoking
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
and drinking status, adherence to Mediterranean Diet, physical activity in leisure time) or
their proxies, e.g. body mass index (BMI). A detailed description of the covariates is
reported in Supplementary Materials. The main pollution exposure was initially tested as
PC scores, as computed through the PCA over all yearly pollutant levels (see above), which
were tested all together in multivariable models. If any PC score showed a significant
association, levels of the single pollutant/s tagged by that PC were tested, averaged over all
the actual years of follow-up for each participant. Sensitivity analyses were carried out
removing all early onset cases (<50 years for PD, <65 years for AD), which are more likely
to be mostly of genetic origins. All the models showed Variance Inflation Factors (VIF) < 2,
suggesting negligible collinearity bias. A Bonferroni correction for multiple testing of three
environmental principal components and two main neurodegenerative outcomes was
applied, resulting in a corrected significance threshold α = 0.008.
Testing the role of circulating inflammation
We tested circulating inflammation as a variable possibly explaining the association
between air pollution and neurodegenerative risk, as hypothesized elsewhere [8,13,30,31].
Specifically, we used a composite blood-based inflammation index, called INFLA-score,
based on four circulating biomarkers - C-reactive protein levels (CRP), blood platelet count
(Plt), white blood cell count (WBC), and granulocyte-to-lymphocyte ratio (GLR) – and
capturing both serum and cellular-circulating inflammation [32]. This score has been
already validated as a comprehensive index of circulating inflammation, since it includes
cytokine-related, hemostatic and immune components of the inflammatory response [33],
and has been previously associated with the inflammatory potential of diet within the Moli-
sani study [34].
For all potential explanatory markers, the analysis was carried out through bootstrap-based
Cox PH regression models (over 1,000 bootstraps). For each bootstrap sampling, we
calculated the change in log (hazard ratio) between the model without and with the
investigated marker, divided by the log (hazard ratio) of the model without the investigated
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
marker. Then, a proportion of total effect explained (PTE) by the marker was computed as
the absolute value of this estimate, and standard deviation (SD) and p-values were defined
using the empirical distribution of coefficients resulting from all bootstraps.
Results
Table 1 reports main sociodemographic and epidemiological characteristics of the analyzed
samples both for AD and PD incident risk. Compared to participants of the Moli-sani cohort
excluded from analyses based on the lack of concordant geolocalization, lack of follow-up
information or prevalent AD/PD condition, analyzed subjects showed a generally higher
education level (p = 0.002 for AD and < 0.0001 for PD analysis) and a lower prevalence of
CVD, cancer, diabetes, hypertension (p < 0.0001 for both AD and PD) and hyperlipidemia
(p = 0.007 for AD, p < 0.0001 for PD), a lower frequency of moderate drinkers and a higher
frequency of heavy/very heavy drinkers (p < 0.0001 for both AD and PD). No significant
difference in sex distribution was observed (48.11% men in the analysis of AD and 48.15%
in the analysis of PD), as well as in all the variables tagging lifestyles, except for leisure time
physical activity, which was higher in the analyzed compared to the removed samples (p =
0.03 for AD and p < 0.0001 for PD), and for BMI, which was slightly lower in the analyzed
sample of the PD analysis (p = 0.03). No significant difference was found in the distribution
of inflammatory markers, except for a marginally lower GLR value in the subjects analyzed
for PD, compared to those removed (Table 1).
A PCA over 24,308 participants with air pollution levels available within the Molise region
revealed three main principal components scores, explaining >5% of common variance
across all environmental variables tested: PC1 (38.2%), PC2 (19.5%) and PC3 (8.5%; Figure
S1). While PC1 was quite clearly tagging PM10 levels, PC2 and PC3 showed moderate to
high loadings of O3/CO/SO2 and NOx/BTX hydrocarbons, although less clearly (Figure S2a,
b, c). This pattern was consistent with the Spearman’s correlation patterns across all the
pollutants tested together, averaged across the years (Figure S3).
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Although Model 3 (adjusted for sociodemographic, professional and lifestyle factors) was
considered as the main model of reference for the interpretation of results, we report below
the results of all the incrementally adjusted Cox PH models to appreciate the influence of
incremental adjustment on the associations analyzed (Table 3). Over 24,195 subjects
analyzed for AD risk (218 incident cases, median (IQR) follow-up 11.17 (2.02) years), we
observed a significant association of PC1 with an increased risk of dementia (HR [95% CI] =
1.06 [1.04-1.08] per unitary increase of PC1 score), which was stable across models adjusted
for professional (1.06 [1.04-1.08]) and lifestyles covariates (1.06 [1.04-1.08]; p-value < 8.7×10-
8). A significant association was also observed for incident Parkinson’s
disease/parkinsonisms risk (N = 23,990, 405 incident cases, median (IQR) follow-up: 11.17
(2.03) years), although with a slightly smaller effect size (1.04 [1.03-1.05]), which was again
confirmed after adjustment for professional covariates and became even larger after
adjusting for lifestyles or their proxies (1.05 [1.03-1.06]; p = 6.6×10-9). No other PC score
showed significant associations surviving Bonferroni correction for multiple testing, with
any of the disorders tested (Table 3), nor any other covariate used in the model, except for
age (Table S2a, b). Since PC1 showed high loadings of PM10 and was therefore clearly
tagging the levels of this pollutant, we tested directly PM10 for association with incident
AD and PD risk, which revealed relative risks consistent with those observed for PC1.
Indeed, each unitary (µg/m3) increase of PM10 was associated with a 25 (19-31)% increase
of incident AD risk and a 19 (15-24)% increase of PD risk in the most conservative model
(Table 3, Model 3). Since PM10 showed a bimodal distribution (Figure S4), to ensure against
potential biases resulting from potential departures from normality, we compared
participants above and below the median level of PM10 in the analyzed population (11.6
µg/m3), which revealed strong associations, in line with the effect sizes observed per unitary
increase. Participants exposed to average PM10 concentrations > 11.6 µg/m3 showed a ⁓22x
(13-39) increase of AD risk and a ⁓14x (10-21) increase of PD risk, compared to subjects
exposed to PM10 concentrations ≤ 11.6 μg/m3 (Figure 2, Table 3). Associations remained
stable after removal of early onset AD and PD cases (Table S3).
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
When we tested potential roles of inflammatory markers like INFLA-score, GLR, CRP, Plt
and WBC, the proportion of the above-mentioned associations explained by these markers
was non-significant and negligible (<1%) for both AD and PD (Table 4).
Discussion
In the present manuscript, we analyzed the relationship between air pollution and incident
neurodegenerative risk in an Italian population, identifying a notable influence of PM10
levels on an increased risk of both dementia forms and Parkinson’s disease/parkinsonisms.
Interestingly, this influence was independent on diverse other factors representing known
risk/protective factors for AD and PD, including lifestyles like physical activity, smoking,
drinking and adherence to Mediterranean Diet [35,36], and on professional factors like
working class and exposure to toxic compounds. While a link between PM10 and
neurodegenerative risk has been long hypothesized, only few studies supported this
through statistical evidence – especially for AD – and recent meta-analyses revealed
contrasting results and a yet unclear relationship [13,15,16]. Indeed, most of the studies
have so far supported an association of increased AD and PD risk with PM2.5 and NO2
[13,15,17,18,37,38], but findings on PM10 are less consistent. A geospatial analysis of PD
cases and controls reported a significant difference in the mean annual NO2 and PM10
levels between areas where PD cases were concentrated (hotspots) and those where they
were not (coldspots) [23]. These findings are in line with a large nested case-control study
from a Chinese health insurance cohort, where PM10 was the only pollutant reported to
influence an increase in PD risk, among many others [21]. However, other cohort studies
found no evidence of association with PM10 [17,20,39,40], as well as case-control studies
[22,41,42], and even meta-analyses [30,43].
Other works identified associations between PM2.5 and clinical events related to these
disorders, like an increased risk of hospitalization for dementia causes (e.g [44]) or an
increased mortality in Parkinson’s disease patients [16]. Some studies like the Rome
Longitudinal Study reported a positive association of both PM2.5 and PM10 with incident
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
hospitalization risk for vascular dementia, but a negative association with Alzheimer
dementia [19], and no association with Parkinson’s disease [20]. Other studies have found
associations between exposure to air pollutants – particularly PM2.5, PM10 and NO2 - and
neuropathological hallmarks of AD and PD [25,45,46,47], neuroimaging endophenotypes
like a reduced prefrontal cortex [48], hippocampal volume [49], and cortical thickness in the
temporal lobe [50].
In spite of these promising findings, the molecular mechanism of action of these pollutants
remains largely unclear. PM represent one of the main suspect to play a role in the
pathophysiology of neurodegeneration since they are small enough to enter the lungs and
spread through the brain, possibly through alteration of the blood brain barrier, oxidative
stress, microglia activation and brain inflammation [8,31,47,51], but also through the
alteration of connectivity among different brain regions [13]. In spite of the above mentioned
role of neuroinflammation in both AD and PD pathophysiology, and of experimental
evidence suggesting that circulating inflammation may somehow mediate the effect of
pollution on neurodegenerative risk [6,24,25], our analysis did not reveal any significant
explanatory effect by these markers, suggesting that other mediating pathways may act in
this link. Among them, biological aging, the actual underlying age of an organism, which
can be measured through different algorithms and biomedical data sources (e.g. epigenetic
variations, blood markers, brain imaging) may represent a key player. Indeed, both long
and short term exposure to air pollution was already reported to influence telomere attrition
and epigenetic aging [52,53,54,55], which very well predict incident risks of dementia and
Parkinson’s disease [56,57]. This represents a promising hypothesis to investigate in the
future, along with gene-by-environment and environment-by-environment interactions
[13,58], which show the potential to further clarify how pollution and other exposome layers
may influence incident neurodegenerative risk.
Of note, one of the pollutants most reported to have an influence on AD and PD risk by
previous studies, NO2 [13], did not reveal any significant association in the present work.
While this may depend on the different approach used here to investigate air pollution
exposure while controlling for collider bias – namely through PC scores rather than single
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
pollutant levels – this finding is not that surprising in view of the inconsistent results
published so far for this and other compounds, which may be due to the same elements of
heterogeneity across studies mentioned above for PM10.
Strengths and Limitations
The present work presents strengths like the comprehensive approach to multiple air
pollutants from diverse sources and the simultaneous assessment of many lifestyles and
professional factors which may influence incident neurodegenerative risk, which allowed
us to identify a clear influence of PM10 on AD and PD, independent on all other risk and
protective factors tested. Moreover, although a role of inflammation has been often
hypothesized, we are not aware of any study testing potential explanations of the link
between pollution and neurodegenerative disorders through a diverse set of circulating
inflammation markers, tagging cytokine-related, hemostatic and cellular/immune
components of inflammation. However, our work also has some limitations. First, the use
of land measurements to interpolate exposure maps may not be as precise as maps
integrating land and satellite data, as well as the fact that data for the first year of
recruitment (2005) were not available in the ARPA Molise database. However, Molise is a
rather rural region which does never experience sudden massive changes in the degree of
anthropic activities, hence the levels of air pollutants tend to remain stable across the years.
This, along with the dimensionality reduction approach used, which was necessary to
reduce collinearity biases, may explain the large effect sizes observed. We plan to further
investigate these associations using environmental data from other independent European
sources, which will be made available in the future. Third, the algorithm used for defining
incident neurodegenerative cases may have led to classify as AD/PD cases subjects affected
by related disorders or with a yet unclear diagnosis, a problem often affecting neurological
clinical practice. However, we remained prudent in the definition of the neurodegenerative
disorders analyzed and still this does not affect the focus of the manuscript, on the
relationship between pollution and incident neurodegenerative risk in the population,
which needs modifiable risk factors to be identified, whatever the clinical diagnosis. Fourth,
circulating inflammation does not necessarily reflect neuroinflammation, which may
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
explain the low proportions of associations explained by circulating inflammatory markers.
Last, the simultaneous assessment of environmental exposures and circulating
inflammation markers does not allow us to infer direction of causality between them, hence
we could not perform a formal mediation analysis. We are now working on improving
many of these aspects, so to run more powerful analyses with more precise exposure
estimates and longer and validated neurological follow-up.
Authors’ contributions
AG, SC, LI, SM and FG, MBD, GdG, CC, GV and MMF contributed to the concept and
design of the work and/or to the interpretation of data. SC, TP, MR and SF managed data
collection, curation and elaboration. AC, FM and GZ carried out elaboration, quality
control and curation of environmental data. AG, SC and AT performed data analysis. AG,
AT, AC, FM and GZ wrote a first draft of the manuscript, with critical contributions from
all the co-authors. All the co-authors approved the final version of the manuscript.
Funding
The enrollment phase of the Moli-sani Study was supported by unrestricted research grants
from the Pfizer Foundation (Rome, Italy), the Italian Ministry of University and Research
(MIUR, Rome, Italy)—Programma Triennale di Ricerca, Decreto no.1588, and
Instrumentation Laboratory, Milan, Italy. The follow-up phase of the Moli-sani Study
(assessment of incident cases) was partially supported the Italian Ministry of Health (PI
GdG, CoPI SC; grant no. RF-2018-12367074). The collection and elaboration of
environmental data was supported by the Italian Ministry of Economic Development
(PLATONE project, bando "Agenda Digitale" PON I&C 2014-2020; Prog. n. F/080032/01-
03/X35). The present analyses were partially supported by the INAIL-Bric 2019 ID 47 project
(Italian National Institute for Insurance against Accidents at Work; call 2019, project CUP
code: F24I19000630008).
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
No funder had a role in study design, collection, analysis, interpretation of data, writing of
the manuscript, and decision to submit this article for publication.
Acknowledgements
The Moli-sani research group thanks the Associazione Cuore Sano Onlus (Campobasso,
Italy) for its cultural support, Marno Srl (Rosignano Marittimo, LI) and Innovation Group
Scrl (Sant’Agapito, IS) for the record linkage with electronic health data.
Data Availability Statement:
The data underlying this article will be shared upon reasonable request to the corresponding
author. The data are stored in an institutional repository (https://repository.neuromed.it)
and access is restricted by the ethics approval and the legislation of the European Union.
References
1. Organización Mundial de la Salud (OMS). WHO global air quality guidelines. Coast
Estuar Process 2021; 1–360.
2. Akimoto H. Global air quality and pollution. Science 2003; 302: 1716–1719.
3. Thurston GD, Kipen H, Annesi-Maesano I et al. A joint ERS/ATS policy statement:
what constitutes an adverse health effect of air pollution? An analytical framework.
Eur Respir J 2017; 49: 1600419.
4. Dockery DW, Pope CA, Xu X et al. An Association between Air Pollution and
Mortality in Six U.S. Cities. https://doi.org/101056/NEJM199312093292401 1993; 329:
1753–1759.
5. Xu X, Ha SU, Basnet R. A Review of Epidemiological Research on Adverse
Neurological Effects of Exposure to Ambient Air Pollution. Front public Heal 2016;
4: 157.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
6. Block ML, Calderón-Garcidueñas L. Air pollution: mechanisms of
neuroinflammation and CNS disease. Trends Neurosci 2009; 32: 506–516.
7. Association A. 2019 Alzheimer ’ s disease facts and figures. Alzheimer’s Dement
2019; 15: 321–387.
8. Choi D, Lee G, Kim KH, Bae H. Particulate Matter Exacerbates the Death of
Dopaminergic Neurons in Parkinson’s Disease through an Inflammatory Response.
Int J Mol Sci 2022; 23.
9. Korczyn AD. Vascular parkinsonism—characteristics, pathogenesis and treatment.
Nat Rev Neurol 2015; 11: 319–326.
10. Kwon DH, Hwang JS, Kim SG, Jang YE, Shin TH, Lee G. Cerebrospinal Fluid
Metabolome in Parkinson’s Disease and Multiple System Atrophy.
International Journal of Molecular Sciences 23 2022.
11. Tan SH, Karri V, Tay NWR et al. Emerging pathways to neurodegeneration:
Dissecting the critical molecular mechanisms in Alzheimer’s disease, Parkinson’s
disease. Biomed Pharmacother 2019; 111: 765–777.
12. Labzin LI, Heneka MT, Latz E. Innate Immunity and Neurodegeneration. Annu Rev
Med 2018; 69: 437–449.
13. Pini L, Salvalaggio A, Wennberg AM, Dimakou A, Matteoli M, Corbetta M. The
pollutome-connectome axis: A putative mechanism to explain pollution effects on
neurodegeneration. Ageing Res Rev 2023; 86: 101867.
14. Palacios N. Air pollution and Parkinson’s disease - Evidence and future directions.
Rev Environ Health 2017; 32: 303–313.
15. Zhao YL, Qu Y, Ou YN, Zhang YR, Tan L, Yu JT. Environmental factors and risks of
cognitive impairment and dementia: A systematic review and meta-analysis. Ageing
Res Rev 2021; 72.
16. Cole-Hunter T, Zhang J, So R et al. Long-term air pollution exposure and Parkinson’s
disease mortality in a large pooled European cohort: An ELAPSE study. Environ Int
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
2023; 171.
17. Jo S, Kim YJ, Park KW et al. Association of NO2and Other Air Pollution Exposures
with the Risk of Parkinson Disease. JAMA Neurol 2021; 78: 800–808.
18. Chen H, Kwong JC, Copes R et al. Exposure to ambient air pollution and the
incidence of dementia: A population-based cohort study. Environ Int 2017; 108: 271–
277.
19. Cerza F, Renzi M, Gariazzo C et al. Long-term exposure to air pollution and
hospitalization for dementia in the Rome longitudinal study. Environ Heal A Glob
Access Sci Source 2019; 18: 1–12.
20. Cerza F, Renzi M, Agabiti N et al. Residential exposure to air pollution and incidence
of Parkinson’s disease in a large metropolitan cohort. Environ Epidemiol 2018; 2.
21. Chen CY, Hung HJ, Chang KH et al. Long-term exposure to air pollution and the
incidence of Parkinson’s disease: A nested case-control study. PLoS One 2017; 12: 1–
14.
22. Toro R, Downward GS, van der Mark M et al. Parkinson’s disease and long-term
exposure to outdoor air pollution: A matched case-control study in the Netherlands.
Environ Int 2019; 129: 28–34.
23. Fleury V, Himsl R, Joost S et al. Geospatial analysis of individual-based Parkinson’s
disease data supports a link with air pollution: A case-control study. Park Relat
Disord 2021; 83: 41–48.
24. Campbell A, Oldham M, Becaria A et al. Particulate Matter in Polluted Air May
Increase Biomarkers of Inflammation in Mouse Brain. Neurotoxicology 2005; 26:
133–140.
25. Levesque S, Taetzsch T, Lull ME et al. Diesel exhaust activates and primes microglia:
Air pollution, neuroinflammation, and regulation of dopaminergic neurotoxicity.
Environ Health Perspect 2011; 119: 1149–1155.
26. Di Castelnuovo A, Costanzo S, Persichillo M et al. Distribution of short and lifetime
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
risks for cardiovascular disease in Italians. Eur J Prev Cardiol 2012; 19: 723–730.
27. Singh P, Tyagi A. Applying kriging approach on pollution data using gis software.
Int J Environ Eng Manag 2013; 4: 185–190.
28. R Core Team. R: A Language and Environment for Statistical Computing. 2020.
29. Therneau TM, Grambsch PM. Modeling survival data : extending the Cox model.
New York: Springer, New York, NY, 2000.
30. Wang Y, Liu Y, Yan H. Effect of long-term particulate matter exposure on
Parkinson’s risk. Environ Geochem Health 2020; 42: 2265–2275.
31. Shabani S. A mechanistic view on the neurotoxic effects of air pollution on central
nervous system: risk for autism and neurodegenerative diseases. Environ Sci Pollut
Res 2021; 28: 6349–6373.
32. Pounis G, Bonaccio M, Di Castelnuovo A et al. Polyphenol intake is associated with
low-grade inflammation, using a novel data analysis from the Moli-sani study.
Thromb Haemost 2016; 115: 344–352.
33. Gialluisi A, Bonaccio M, Di Castelnuovo A et al. Lifestyle and biological factors
influence the relationship between mental health and low-grade inflammation. Brain
Behav Immun 2020; 85: 4–13.
34. Shivappa N, Bonaccio M, Hebert JR et al. Association of proinflammatory diet with
low-grade inflammation: results from the Moli-sani study. Nutrition 2018; 54: 182–
188.
35. Belvisi D, Pellicciari R, Fabbrini G, Tinazzi M, Berardelli A, Defazio G. Modifiable
risk and protective factors in disease development, progression and clinical subtypes
of Parkinson’s disease: What do prospective studies suggest? Neurobiol Dis 2020;
134: 104671.
36. Dukelow T, Lawrence EG, Jacobson L et al. Modifiable risk factors for dementia, and
awareness of brain health behaviors: Results from the Five Lives Brain Health
Ireland Survey (FLBHIS). Front Psychol 2023; 13.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
37. Mortamais M, Gutierrez LA, de Hoogh K et al. Long-term exposure to ambient air
pollution and risk of dementia: Results of the prospective Three-City Study. Environ
Int 2021; 148: 106376.
38. Dhiman V, Trushna T, Raj D, Tiwari RR. Is ambient air pollution a risk factor for
Parkinson’s disease? A meta-analysis of epidemiological evidence. Int J Environ
Health Res 2022; 00: 1–18.
39. Palacios N, Fitzgerald KC, Hart JE et al. Air pollution and risk of parkinson’s disease
in a large prospective study of men. Environ Health Perspect 2017; 125: 1–7.
40. Palacios N, Fitzgerald KC, Hart JE et al. Particulate matter and risk of Parkinson
disease in a large prospective study of women. Environ Heal A Glob Access Sci
Source 2014; 13: 1–9.
41. Liu R, Young MT, Chen JC, Kaufman JD, Chen H. Ambient Air Pollution Exposures
and Risk of Parkinson Disease. Environ Health Perspect 2016; 124: 1759–1765.
42. Lee PC, Liu LL, Sun Y et al. Traffic-related air pollution increased the risk of
Parkinson’s disease in Taiwan: A nationwide study. Environ Int 2016; 96: 75–81.
43. Han C, Lu Y, Cheng H, Wang C, Chan P. The impact of long-term exposure to
ambient air pollution and second-hand smoke on the onset of Parkinson disease: a
review and meta-analysis. Public Health 2020; 179: 100–110.
44. Kioumourtzoglou M-A, Schwartz JD, Weisskopf MG et al. Long-term PM2.5
Exposure and Neurological Hospital Admissions in the Northeastern United States.
Environ Health Perspect 2016; 124: 23–29.
45. Alemany S, Crous-Bou M, Vilor-Tejedor N et al. Associations between air pollution
and biomarkers of Alzheimer’s disease in cognitively unimpaired individuals.
Environ Int 2021; 157: 106864.
46. Iaccarino L, La Joie R, Lesman-Segev OH et al. Association Between Ambient Air
Pollution and Amyloid Positron Emission Tomography Positivity in Older Adults
With Cognitive Impairment. JAMA Neurol 2021; 78: 197–207.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
47. Calderón-Garcidueñas L, Solt AC, Henríquez-Roldán C et al. Long-term Air
Pollution Exposure Is Associated with Neuroinflammation, an Altered Innate
Immune Response, Disruption of the Blood-Brain Barrier, Ultrafine Particulate
Deposition, and Accumulation of Amyloid β-42 and α-Synuclein in Children and
Young Adults. http://dx.doi.org/101177/0192623307313011 2008; 36: 289–310.
48. Gale SD, Erickson LD, Anderson JE, Brown BL, Hedges DW. Association between
exposure to air pollution and prefrontal cortical volume in adults: A cross-sectional
study from the UK biobank. Environ Res 2020; 185: 109365.
49. Hedges DW, Erickson LD, Kunzelman J, Brown BL, Gale SD. Association between
exposure to air pollution and hippocampal volume in adults in the UK Biobank.
Neurotoxicology 2019; 74: 108–120.
50. Cho J, Noh Y, Kim SY et al. Long-Term Ambient Air Pollution Exposures and Brain
Imaging Markers in Korean Adults: The Environmental Pollution-Induced
Neurological EFfects (EPINEF) Study. Environ Health Perspect 2020; 128: 1–11.
51. Milani C, Farina F, Botto L et al. Systemic Exposure to Air Pollution Induces
Oxidative Stress and Inflammation in Mouse Brain, Contributing to
Neurodegeneration Onset. International Journal of Molecular Sciences 21 2020.
52. Miri M, Nazarzadeh M, Alahabadi A et al. Air pollution and telomere length in
adults: A systematic review and meta-analysis of observational studies. Environ
Pollut 2019; 244: 636–647.
53. Gao X, Huang J, Cardenas A et al. Short-Term Exposure of PM2.5and Epigenetic
Aging: A Quasi-Experimental Study. Environ Sci Technol 2022; 56: 14690–14700.
54. Baranyi G, Deary IJ, McCartney DL et al. Life-course exposure to air pollution and
biological ageing in the Lothian Birth Cohort 1936. Environ Int 2022; 169: 107501.
55. Zizza A, Panico A, Grassi T et al. Is telomere length in buccal or salivary cells a
useful biomarker of exposure to air pollution? A review. Mutat Res Toxicol Environ
Mutagen 2022; 883–884: 503561.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
56. Fransquet PD, Lacaze P, Saffery R et al. Accelerated Epigenetic Aging in Peripheral
Blood does not Predict Dementia Risk. Curr Alzheimer Res 2021; 18: 443–451.
57. Degerman S, Josefsson M, Nordin Adolfsson A et al. Maintained memory in aging is
associated with young epigenetic age. Neurobiol Aging 2017; 55: 167–171.
58. Zhang Z, Chen L, Wang X et al. Associations of Air Pollution and Genetic Risk With
Incident Dementia: A Prospective Cohort Study. Am J Epidemiol 2023; 192: 182–194.
59. Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to a
Mediterranean Diet and Survival in a Greek Population. N Engl J Med 2003; 348:
2599–2608.
URLs
ARPA Molise: https://www.arpamolise.it/
Kriging algorithm: https://desktop.arcgis.com/en/arcmap/10.3/tools/3d-analyst-
toolbox/how-kriging-works.htm
GeoKettle: http://www.geokettle.org/
HERE: https://www.here.com/
QGis: https://www.qgis.org/en/site/
R: https://www.r-project.org/
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Table 1. Baseline characteristics of the population under study.
Variable
Population under
study for AD
(N = 24,195)
Removed
participants
(N = 130)
P for difference
(analyzed vs removed)
Population under
study for PD
(N = 23,990)
Removed
participants
(N = 335)
P for difference
(analyzed vs
removed)
Sex (men, %)
48.11
47.69
0.93
48.15
45.37
0.32
Age (y)
55.76 (11.94)
61.64 (14.61)
<.0001
55.69 (11.92)
62.94 (12.97)
<.0001
Education (%)
Primary or less
Lower secondary
Upper secondary
Post-secondary
Missing
25.78
27.79
34.04
12.38
-
23.08
13.85
16.92
16.15
30.00
0.002
25.61
27.83
34.14
12.43
-
37.31
19.40
20.90
10.75
11.64
<.0001
Health conditions (%)
CVD
Cancer
Diabetes
Hypertension
Hyperlipidaemia
5.89
3.50
4.98
28.70
7.89
14.96
5.77
6.92
32.31
15.08
<.0001
<.0001
<.0001
<.0001
0.007
5.81
3.49
4.94
28.56
7.82
15.24
3.56
8.96
40.30
16.36
<.0001
<.0001
<.0001
<.0001
<.0001
Lifestyle factors
Smokers (%)
MeDi Score (0-9)
Physical activity (MET-h/day)
BMI (kg/m2)
Moderate alcohol drinkers (%)
Heavy/very heavy alcohol drinkers
(%)
22.99
4.35 (1.64)
3.49 (4.02)
28.06 (4.78)
15.16
27.87
20.34
4.24 (1.66)
3.00 (3.68)
28.00 (5.43)
21.43
21.43
0.59
0.47
0.03
0.90
<.0001
23.01
4.35 (1.64)
3.49 (4.02)
28.05 (4.77)
15.15
27.89
20.12
4.18 (1.61)
2.97 (4.05)
28.66 (5.07)
18.37
24.02
0.41
0.06
<.0001
0.03
<.0001
Systemic Inflammation
INFLA-score
CRP (mg/L)
WBC (×109/L)
GLR
-0.09 (6.03)
2.59 (3.26)
6.23 (1.75)
2.02 (0.93)
0.33 (6.03)
2.76 (3.65)
6.70 (4.38)
2.10 (0.92)
0.43
0.65
0.17
0.32
-0.10 (6.03)
2.59 (3.26)
6.23 (1.76)
2.01 (0.93)
0.45 (6.05)
2.91 (3.53)
6.35 (3.00)
2.12 (0.86)
0.11
0.11
0.77
0.02
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Here we report frequency (%) for categorical variables, or, alternatively, mean values and standard deviations (SD) for continuous
variables. P-values resulting from statistical comparisons of the analyzed vs non-analyzed participants both for AD and for PD analysis
are reported. Chi-squared test was applied to education levels, smoking and alcohol drinking classes; Fisher Exact Test to sex, CVD,
cancer, diabetes, hypertension and hyperlipidemia; unpaired t-test to age, INFLA-score, GLR, MeDi score and BMI; and Wilcoxon signed
rank test to CRP, WBC and physical activity levels. All the variables are defined in Supplementary Materials. Abbreviations: BMI= body
mass index; CVD = cardiovascular disease; MeDi = adherence score to Mediterranean Diet (MeDi)[59]; CRP = C-reactive protein; WBC
= white blood cells count; GLR = granulocyte-to-lymphocite ratio.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Table 2. Results of the main analysis of incident a) AD and b) PD risk vs principal components of air pollutants exposure.
a)
HR [95% CI]
Model 1
HR [95% CI]
Model 2
HR [95% CI]
Model 3
p
Model 3
Exposure
Tagged pollutants
1.06 [1.04-1.08]
1.06 [1.04-1.08]
1.06 [1.04-1.08]
8.7×10-8
PC1
PM10
0.97 [0.94-1.00]
0.96 [0.93-0.99]
0.96 [0.93-0.99]
0.014
PC2
O3/CO/SO2
0.99 [0.95-1.04]
0.99 [0.95-1.04]
1.00 [0.96-1.05]
0.87
PC3
NOx/BTX hydrocarbons
b)
HR [95% CI]
Model 1
HR [95% CI]
Model 2
HR [95% CI]
Model 3
p
Model 3
Exposure
Tagged pollutants
1.04 [1.03-1.05]
1.04 [1.03-1.06]
1.05 [1.03-1.06]
6.6×10-9
PC1
PM10
1.00 [0.98-1.02]
1.00 [0.98-1.02]
1.01 [0.98-1.03]
0.54
PC2
O3/CO/SO2
0.98 [0.95-1.01]
0.98 [0.96-1.01]
0.99 [0.96-1.02]
0.63
PC3
NOx/BTX hydrocarbons
Hazard Ratios and 95% confidence intervals HR [95% CI] are reported for each model and each PC score tested, while association p-
value is reported only for the most adjusted model (see below). Significant associations surviving Bonferroni correction for multiple
testing (α = 8×10-3) are highlighted in bold. Model 1: age, sex and education level completed; Model 2: Model 1 + professional exposure
to toxic compounds and working class; Model 3: Model 2 + lifestyles + BMI. Legend: PM10 = particulate matter with aerodynamic
diameter < 10; NOx = nitrogen oxides; SO2 = sulfur dioxide; CO = Carbon monoxide; O3 = ozone; BTX hydrocarbons = benzene, toluene
and xylene.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Table 3. Results of the main analysis of incident AD and PD risk vs PM10 levels.
HR [95% CI]
Model 1
HR [95% CI]
Model 2
HR [95% CI]
Model 3
p
Model 3
Outcome
Exposure
1.22 [1.18-1.28]
1.24 [1.18-1.29]
1.25 [1.19-1.31]
7.5×10-20
AD
PM10 (µg/m3)
-
-
22.12 [12.51-39.14]
1.9×10-26
AD
PM10 (high vs low exposure)
1.18 [1.14-1.21]
1.19 [1.15-1.22]
1.19 [1.15-1.24]
1.5×10-23
PD
PM10 (µg/m3)
-
-
14.24 [9.90-20.47]
1.5×10-46
PD
PM10 (high vs low exposure)
Hazard Ratios and 95% confidence intervals (HR [95% CI]) are reported for each model, while association p-value is reported only for
the most adjusted model (see below). Model 1: age, sex and education level completed; Model 2: Model 1 + professional exposure to
toxic compounds and working class; Model 3: Model 2 + lifestyles + BMI. Both HR associated with unitary increase of PM10 (µg/m3) and
for participants exposed to high vs low exposure compared to median PM10 levels (11.6 µg/m3) are reported. Legend: AD = Alzheimer’s
Disease/Dementia; PD = and Parkinson's Disease/Parkinsonisms; PM10 = particulate matter with aerodynamic diameter < 10 μm.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Table 4. Proportion of association between air pollution (PM10) and incident AD/ PD risk explained by circulating inflammation.
Outcome
INFLA-score
PTE (SD),
p-value
CRP
PTE (SD),
p-value
Plt
PTE (SD),
p-value
WBC
PTE (SD),
p-value
GLR
PTE (SD),
p-value
AD
0.59 (0.53)%
p = 0.43
0.57 (0.49)%
p = 0.42
0.57 (0.52)%
p = 0.42
0.61 (0.55)%
p = 0.45
0.59 (0.55)%
p = 0.42
PD
0.67 (0.61)%
p = 0.71
0.71 (0.64)%
p = 0.74
0.66 (0.65)%
p = 0.70
0.70 (0.60)%
p = 0.74
0.68 (0.63)%
p = 0.72
We report percentage of total effect (PTE) explained by the INFLA-score [7] and its component biomarkers in the association between
PM10 levels and incident AD and PD risk, along with relevant standard deviation (SD) and empirical p-value, as computed through
bootstrapping simulations. These estimates refer to Model 3, adjusted for sociodemographic, professional and lifestyle covariates.
Abbreviations: AD = dementia; PD : Parkinson’s disease/parkinsonisms; CRP = C-reactive protein; Plt = platelets count; GLR =
granulocyte-to-lymphocite ratio; WBC = white blood cells count.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Figure 1. Model used to build air pollutants exposure maps in the Moli-sani cohort.
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
Figure 2. Kaplan Meier curves of incident a) AD and b) PD events vs PM10 quantiles.
Incident a) dementia and b) Parkinson’s disease/parkinsonisms events are compared
between the two quantiles of exposure to PM10: below the median (11.6 μg/m3; green)
and above the median (red)
All rights reserved. No reuse allowed without permission.
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprintthis version posted May 18, 2023. ; https://doi.org/10.1101/2023.05.17.23289154doi: medRxiv preprint
