Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Nov 3.
Published in final edited form as: Sci Total Environ. 2026 Jun 30;1046:181999. doi: 10.1016/j.scitotenv.2026.181999

Traffic-related air toxics and Parkinson’s disease in California’s Central Valley

Dayoon Kwon a, Kanghong Shao a, Karl O’Sharkey b, Yu Yu b,c, Jason Su d, Jeff M Bronstein e, Beate Ritz b,e
PMCID: PMC13727049  NIHMSID: NIHMS2217038  PMID: 42378886

Abstract

Growing evidence links air pollution to Parkinson’s disease (PD), yet most research focuses on criteria pollutants like nitrogen dioxide (NO2) and fine particulate matter (PM2.5), which cannot pinpoint specific toxic constituents driving the association. This population-based case-control study from the counties of Fresno, Tulare, and Kern in central California (764 patients, 798 controls) investigated whether specific air toxics – metals with mobile-source/industrial contributions (nickel, zinc), vehicle-exhaust volatile organic compounds (benzene, 1,3-butadiene), and industrial/urban metals (lead, chromium) – are associated with PD. Using land-use regression models linked to participant’s full residential and workplace address histories, we estimated 10-year, 5-year, and 1-year average pollutant exposures preceding diagnosis (patients) or interview year (controls). Associations were assessed via logistic regression, adjusting for sociodemographics and pesticide exposure. For nickel we estimated the strongest and most consistent effect across all averaging periods, with a 10-year residential odds of 1.63 (95% CI = 1.04–2.55) and a nearly three-fold increase in the odds of PD due to ambient exposure at the workplace (OR = 2.89, 95% CI = 1.48–5.80) per interquartile range increase. Zinc showed a positive 10-year residential association (OR = 1.42, 95% CI = 0.98–2.10). The volatile organic compounds benzene (5-year workplace OR = 1.83, 95% CI = 1.16–2.93) and 1,3-butadiene (1-year workplace OR = 4.56, 95% CI = 2.86–7.49) exhibited stronger associations at shorter averaging windows, particularly at workplace addresses. In co-pollutant models, air toxics associations persisted after adjustment for NO2 or PM2.5, while criteria pollutant associations attenuated. No associations were observed for lead or chromium in these predominantly agricultural counties. Our findings highlight that traffic-related and industrial air toxics from both non-exhaust and exhaust sources may contribute to PD risk, underscoring the importance of evaluating pollutant sources and components when assessing environmental determinants of PD.

Keywords: Parkinson’s disease, Air pollution, Air toxics, Traffic-related air pollution

1. Introduction

Parkinson’s disease (PD) is now the fastest-growing neurological disorder worldwide, a trend that cannot be explained by population aging alone (Dorsey et al., 2018a). Incidence rates are increasing beyond demographic expectations (Dorsey et al., 2018b), pointing to widespread environmental drivers. Identifying modifiable risk factors is therefore important, as current treatments remain symptomatic and do not slow disease progression.

Air pollution has emerged as a candidate environmental risk factors for PD. The State of Global Air report included neurodegeneration for the first time in 2025, quantifying the burden of dementia attributable to air pollution (Health Effects Institute, 2025), though PD has yet to be incorporated. Most epidemiological research on air pollution and neurodegeneration has focused on criteria pollutants such as fine particulate matter (PM2.5), PM10, nitrogen dioxide (NO2), nitrogen oxides (NOx), and ozone (Atterling Brolin et al., 2025; Krzyzanowski et al., 2024), yet these aggregate measures cannot distinguish the specific toxic constituents from exhaust or brake and tire wear that may be driving the observed associations. Regulatory actions have reduced tailpipe emissions, but traffic-related PM2.5 components have remained high, and concerns are mounting that unregulated or unmeasured components of traffic emissions may carry distinct neurotoxic risks (Fussell et al., 2022). Despite toxicological evidence implicating certain metals (Anyachor et al., 2022; Shvachiy et al., 2023; Sikora and Ouagazzal, 2021; Wise et al., 2022) and volatile organic compounds (VOCs) (Holland et al., 2022; Zheng et al., 2024) as harmful to brain health, only a small number of epidemiologic studies have evaluated air toxics in relation to neurological outcomes, including autism spectrum disorder (O’Sharkey et al., 2025), dementia and cognitive impairment (Yu et al., 2025), and PD (Palacios et al., 2014).

The need for such investigation is especially pressing in California’s Central Valley, which remains the region in the United States most impacted by PM2.5 pollution (American Lung Association, 2025) due to pollutant trapping by topography, extensive agricultural and some industrial activity, and heavy traffic. These regional features together create large, spatially structured exposure contrasts in traffic-related air toxics that are well-suited to investigation in a population-based case-control design. We therefore investigated whether exposures to six air toxics – nickel and zinc (metals with mobile-source and industrial contributions), benzene and 1,3-butadiene (vehicle-exhaust VOCs), and lead and chromium (industrial and urban sources) – are associated with PD risk at residential and workplace addresses in central California. We compared per-interquartile range (IQR) effect estimates for each air toxics with those for NO2 and PM2.5 in single-pollutant models, and further fit two-pollutant models for each air toxic and one of the criteria pollutants to assess whether the air toxics associations were independent of these two criteria pollutants.

2. Methods

2.1. Study population

The Parkinson’s Environment and Genes (PEG) study is a population-based case-control study conducted in three counties of central California: Fresno, Tulare, and Kern (Kang et al., 2005; Wang et al., 2011). Participants were recruited in two waves (PEG1: 2000–2007; PEG2: 2010–2017). PEG1 participants were recruited from multiple sources, including local neurologist offices, PD support groups, local media, and public radio. PEG2 participants were recruited as part of a population-based PD registry pilot program, in addition to the previously utilized sources listed above.

Patients were eligible if they were early in their disease at enrollment (<5 years since diagnosis, on average 3 years), had their PD diagnosis confirmed by a UCLA movement disorder specialist, resided in California for at least five years prior to diagnosis, had no other severe neurological or terminal health conditions, and agreed to participate in the study. Population controls were recruited using several strategies to ensure representativeness and comparability. In PEG1, we randomly sampled residential parcels from county tax-assessor records and Medicare list, contacted households via mail and telephone, and enrolled eligible residents. In PEG2, we expanded coverage using door-to-door recruitment: field staff visited clusters of five neighboring households randomly selected from county tax-assessor records to enroll one eligible adult per household. Controls followed the same eligibility criteria as cases, except for PD diagnosis.

The final analytic sample included 764 PD patients and 798 controls who provided complete lifetime home and work address histories (Figure S1). Demographic, lifestyle, and medical history data were collected from each participant via a standardized interview conducted by trained PEG study staff. Pesticide exposures were estimated using a geographic information system (GIS)-based model linking California Pesticide Use Reporting (PUR) records to participants’ geocoded home and work addresses (Cockburn et al., 2011; McElroy et al., 2003). For each pesticide and participant, we calculated annual pounds applied per acre within a 500m buffer of each address, adjusting for the portion of acreage treated (lbs/acre) (Paul et al., 2023) (Supplemental Methods). All participants provided informed consent, and the study was approved by the UCLA Institutional Review Board.

2.2. Air pollution exposure assessment

We estimated exposures to six air toxics (nickel, zinc, benzene, 1,3-butadiene, lead, chromium) and two criteria air pollutants (NO2, PM2.5) for all participants from 1989–2019 using land use regression (LUR) models built with a deletion/substitution/addition (D/S/A) machine-learning framework at 100m spatial resolution (Supplemental Methods). LUR was selected for its well-known ability to generate fine-scale spatial gradients based on predictors such as land use and air pollution monitoring data; the D/S/A framework accommodates both linear and non-linear predictor-pollutant relationships. We selected these pollutants based on (1) consistent statewide monitoring by CARB since the 1990s, (2) availability of ≥46 monitoring sites to support stable LUR modeling, and (3) relevance to traffic and urban/industrial emissions in the study region (O’Sharkey et al., 2025; Yu et al., 2025). Participants’ residential and occupational addresses were geocoded (McElroy et al., 2003), and LUR-predicted concentration were extracted at each geocoded address by point-in-raster sampling at the native resolution. Details on monitoring coverage and model performance (such as adjusted R2) are provided in Tables S1–S2.

2.2.1. Air toxics

Air toxics refer to hazardous air pollutants identified by the US Environmental Protection Agency and California Air Resources Board (CARB). We estimated annual concentrations of each air toxic using LUR models that incorporated CARB speciation monitoring data and emissions data from the California Toxics Inventory, including point sources reported under the Air Toxics “Hot Spots” Program (AB2588) and aggregated emissions from local air districts. Predictors included traffic data, land-use and land-cover characteristics, meteorological variables (GridMET), vegetation indices (NDVI), Sentinel-2 spectral bands, and satellite-derived aerosol optical depth (AOD) and trace gas products. The source designations below reflect the plausible dominant sources implicated by each pollutant’s predictor structure.

The six air toxics evaluated here span a range of source profiles (O’Sharkey et al., 2025; Yu et al., 2025). LUR modeling included various sources that contribute to the concentrations of the measured pollutants. While roadway traffic is a significant contributor to the spatial variability of several air toxics, including nickel and zinc, the modeled concentrations represent integrated exposures from multiple potential sources, including traffic emissions, industrial activities, soil and dust resuspension, and agricultural operations. Nickel is emitted from heavy-duty diesel combustion and from metal-containing brake and tire wear (Fussell et al., 2022; Lopez et al., 2023); its LUR surface receives a positive contribution from proximity to goods-movement corridors (highways allowing truck traffic and major ports), alongside urban/developed-land predictors, consistent with a heavy-duty-diesel-weighted mobile-source interpretation. Zinc is a non-exhaust traffic tracer reflecting tire wear and brake abrasion (Councell et al., 2004; Degaffe and Turner, 2011), but its LUR surface is predicted primarily by proximity to industrial land and only secondarily by roadway area; consistent with a mixed-source exposure that includes non-exhaust traffic, industrial, leaching from galvanized infrastructure, and crustal contributions.

Benzene and 1,3-butadiene are VOCs released primarily from motor vehicle exhaust and combustion-related processes (Holland et al., 2022); the benzene surface is predicted by roadway area together with residential, commercial, and developed-land predictors, and the 1,3-butadiene surface by vehicle-kilometers traveled together with industrial and reactive-organic-gas predictors, supporting their role as mobile-source-dominant markers.

Chromium and lead reflect industrial and urban sources: chromium is emitted from metal processing operations (Wise et al., 2022), and lead remains relevant due to its environmental persistence in soil with legacy contamination from leaded fuels despite the decades old phase-out of leaded gasoline and also due to ongoing industrial emissions (Chen et al., 2025; Raj and Das, 2023).

2.2.2. Criteria air pollutants

Criteria pollutants are ambient pollutants regulated under US Clean Air Act. NO2 and PM2.5, serve as our comparators because they are criteria pollutants most consistently linked to PD in the prior literature (Atterling Brolin et al., 2025; Krzyzanowski et al., 2024). NO2 is widely used as a marker of traffic-related air pollution (Boogaard et al., 2022), particularly on-road mobile sources, although in the Central Valley it also receives contributions from off-road mobile sources (agricultural and construction equipment), stationary combustion, and industrial point sources. PM2.5 reflects total ambient fine particulate matter from multiple sources – including mobile sources, residential and agricultural burning, secondary inorganic and organic aerosol formation, and industrial emissions – rather than traffic emissions alone (Hopke et al., 2020).

Daily NO2 concentrations were estimated using fixed-site regulatory monitoring, research-based saturation sampling, and Google Street View mobile measurements, while PM2.5 estimates relied on regulatory monitors and mobile measurements (Su et al., 2024). Models incorporated daily traffic activity (vehicle miles traveled), satellite-based NO2, and AOD products, biweekly vegetation indices, meteorological variables, and land-use/land-cover data.

2.2.3. Long-term exposure assessment

To assess long-term air pollution exposure, we computed 10-year average exposure concentrations prior to the index year, defined as the year of PD diagnosis for patients and the interview year for controls. To ensure temporal comparability between patients and controls, we aligned index years within each study wave (Figure S2). For PEG1, PD patients were enrolled up to 3 years after diagnosis (mean PD duration at baseline: 1.9 years); thus, we adjusted the control index year by subtracting this average duration to generate a calendar time aligned exposure window. As a result, the mean index year for PD patients was 2002.1 (SD = 2.5 years), compared to 2002.5 (SD = 1.6 years) for controls after adjustment. For PEG2, patients and controls were recruited more contemporaneously in terms of the diagnosis year of PD, and the mean index calendar years were well-aligned as 2009.1 (SD = 3.1 years) for patients and 2009.2 (SD = 0.5 years) for controls without additional alignment. For participants missing air pollution data for up to 40% of the 10-year averaging period (4 patients and 3 controls), we imputed missing years using the mean of adjacent years (Table S3).

2.3. Statistical analysis

Using unconditional logistic regression models, we generated effect estimates (odds ratios [ORs] and 95% confidence intervals [CIs]) separately for residential and occupational addresses. Air pollutant exposures were analyzed as continuous variables (per IQR). To ensure comparability of effect estimates across locations, the IQR was calculated from the combined distribution of residential and workplace exposures rather than separately for each. Model 1 was adjusted for age (years), gender, race (white vs. non-white), smoking status (never, former, current), education level (years), family history of PD, study wave (PEG1, PEG2), and index year. Model 2 additionally adjusted for pesticide exposure, an established risk factor for PD in this agricultural population, defined as the average pounds of all pesticides applied per acre within a 500m buffer of the residential or occupational address during the same exposure window (Supplemental Methods). This adjustment is intended to assess associations with these air pollutants adjusted for co-exposures from agricultural activity.

In sensitivity analyses, we investigated two pollutant co-pollutant models to explore the impact of potential confounding and to avoid potential collinearity generated by inclusion of multiple correlated pollutants. Additionally, we evaluated shorter exposure periods (including 5-year and 1-year prior to PD diagnosis for patients and interview date for controls) to assess potential variability in the timing of exposure on PD risk. We also conducted sensitivity analyses restricted to participants for whom we had both residential and workplace addresses during the 10-year exposure window (i.e., they had address data in both domains), to ensure that both effect estimates were based on the same individuals. To reduce exposure misclassification in the pesticide covariate, we repeated analyses using a more restricted pesticide exposure measure that reflects 26 active ingredients (Paul et al., 2023) previously identified as PD-associated at FDR<0.01 in our pesticide-wide association study (PWAS; see Supplemental Methods). To assess whether associations for shorter windows may be affected by behavioral changes (e.g., reduced mobility, retirement) in the years immediately preceding diagnosis, we additionally fit a 1-year lagged sensitivity analysis (a 1-year average converging the period from 2 years to 1 year before the index date, excluding the year immediately preceding diagnosis or interview).

All statistical analyses were performed using R, version 4.3.2 (R Project for Statistical Computing).

3. Results

The mean age at diagnosis for PD patients was 68 years, while the controls had a mean age of 66 years at their index date (Table 1). A higher proportion of patients were male (64%) compared to controls (47%), and most participants were white (76% patients, 69% controls). Education levels were similar, with both groups averaging about 14 years of education. Among the 10-year air toxics, we found high correlations between LUR-modeled traffic-related air toxics, including nickel, benzene, and 1,3-butadiene (ρ ≥ 0.64; Figure S3). In contrast, zinc showed inverse correlations with these traffic-related air toxics and NO2 (ρ = −0.51 to −0.83), likely because its LUR surface captures additional spatial features (e.g., vegetation/open space) that elevate predicted levels away from major roadways and human residences (Figure 1), while near roadways and human residences zinc serves primarily as a marker of non-tailpipe traffic-related metal mixtures (e.g., tire/brake wear components). Correlations between NO2 and traffic-related air toxics ranged from moderate to strong, including nickel (ρ = 0.71), benzene (ρ = 0.66), and 1,3-butadiene (ρ = 0.51), reflecting their shared traffic-related sources. Correlations between the remaining air pollutants were moderate.

Table 1.

Demographic characteristics of study participants.

Characteristica PD patients
(n=764)		 Controls
(n=798)
Age, years 67.9 (10.6) 66.1 (11.6)
Gender
 Male 486 (63.6%) 372 (46.6%)
 Female 278 (36.4%) 426 (53.4%)
Race and ethnicity
 White 583 (76.3%) 552 (69.2%)
 Latino 133 (17.4%) 152 (19.0%)
 Native American 23 (3.0%) 42 (5.3%)
 Asian 20 (2.6%) 25 (3.1%)
 African American 5 (0.7%) 27 (3.4%)
Education, years 13.6 (4.49) 14.0 (3.94)
Smoking status
 Never 408 (53.4%) 383 (48.0%)
 Former 327 (42.8%) 323 (40.5%)
 Current 29 (3.8%) 92 (11.5%)
Study wave
 PEG1 315 (41.2%) 384 (48.1%)
 PEG2 449 (58.8%) 414 (51.9%)
PD family history
 No 594 (77.7%) 733 (91.9%)
 Yes 170 (22.3%) 65 (8.1%)
Air pollution-related job history b
 No 588 (77.0%) 675 (84.6%)
 Yes 176 (23.0%) 123 (15.4%)
Open in a new tab
a

Data are n (%) or mean (SD).

b

Includes any history of employment as a traffic police officer, gas station attendant, mechanic, asphalt worker, or a professional truck, tractor, or bus driver.

PD=Parkinson’s disease.

Figure 1:

Figure 1:

Open in a new tab

Estimated annual average concentrations of air toxics in central California in 1990, generated using land-use regression models.

3.1. Traffic-related pollutants

Unless otherwise specified, effect estimates reported in-text refer to Model 2 (fully adjusted). Among air toxics modeled based on traffic-related data, nickel exhibited the strongest and most consistent associations with PD. The 10-year average exposure at the home address was associated with roughly 1.6-fold higher odds of PD per IQR increment (OR = 1.63, 95% CI = 1.04–2.55; Table 2), and the increase from ambient workplace exposure were nearly three-fold (OR = 2.89, 95% CI = 1.48–5.80). Zinc followed a broadly similar pattern, with positive effect estimates for exposure at residential-addresses (OR = 1.42, 95% CI = 0.98–2.10), and similar but slightly weaker workplace-address based estimates (OR = 1.32, 95% CI = 0.79–2.23) and 95% CIs crossing the null. While the associations for these particulate-phase metals were relatively stable across all three averaging periods (Figure 2), the VOCs that indicate exhaust tended to show stronger signals at shorter exposure windows. For benzene, the 5-year residential-address window showed a positive association (OR = 1.42, 95% CI = 1.08–1.89), and shorter-term workplace-address based estimates were even stronger (5-year OR = 1.83, 95% CI = 1.16–2.93; 1-year OR = 1.73, 95% CI = 1.05–2.85). 1,3-butadiene showed a distinctive pattern, with the 10-year residential-address estimate suggesting weakly elevated odds (OR = 1.12, 95% CI = 0.93–1.35), while 1-year estimates were extremely high with approximately four-fold higher odds at both the residential (OR = 4.02, 95% CI = 3.03–5.40) and workplace addresses (OR = 4.56, 95% CI = 2.86–7.49).

Table 2.

Association between 10-year average air pollution exposure and Parkinson’s disease.

OR (95% CI)
Residence
(764 cases/798 controls)		 Workplace
(342 cases/340 controls)
Air toxica IQR Model 1b Model 2c Model 1b Model 2c
Nickel 0.28 ng/m3 1.62 (1.04–2.54) 1.63 (1.04–2.55) 2.66 (1.37–5.27) 2.89 (1.48–5.80)
Zinc 0.79 ng/m3 1.42 (0.97–2.10) 1.42 (0.98–2.10) 1.27 (0.77–2.15) 1.32 (0.79–2.23)
Benzene 0.22 ppb 1.14 (0.90–1.43) 1.14 (0.90–1.44) 1.26 (0.88–1.81) 1.35 (0.94–1.96)
1,3-Butadiene 0.33 ppb 1.12 (0.93–1.35) 1.12 (0.93–1.35) 1.38 (1.06–1.81) 1.40 (1.07–1.84)
Lead 0.59 ng/m3 0.96 (0.83–1.10) 0.96 (0.83–1.10) 1.03 (0.93–1.14) 1.03 (0.93–1.15)
Chromium 1.47 ng/m3 1.05 (0.90–1.22) 1.06 (0.90–1.24) 1.01 (0.89–1.16) 1.04 (0.91–1.19)
Criteria pollutant a
NO2 2.82 μg/m3 1.06 (0.92–1.23) 1.06 (0.92–1.23) 1.23 (1.02–1.48) 1.27 (1.05–1.54)
PM2.5 1.76 μg/m3 1.06 (0.95–1.17) 1.06 (0.95–1.17) 1.18 (1.00–1.39) 1.21 (1.03–1.43)
Open in a new tab
a

Each pollutant was entered into the model separately. The reported ORs are derived from single-pollutant models.

b

Adjusted for age, race, gender, smoking status, education, family history of PD, study wave, and index year.

c

Additionally adjusted for pesticide exposure.

OR=odds ratio. CI=confidence interval. IQR=interquartile range. NO2, = nitrogen dioxide. PM2.5 = particulate matter ≤2.5 μm.

Figure 2:

Figure 2:

Open in a new tab

Associations by averaging window (10-year primary, 5-year, 1-year) for air pollution and Parkinson’s disease. Odds ratios (ORs) and 95% confidence intervals (95% CIs) per interquartile range (IQR). Circles indicate residential and triangles workplace. Adjusted for age, race, gender, smoking status, education, family history of PD, study wave, index year, and pesticide exposure. NO2, = nitrogen dioxide. PM2.5 = particulate matter ≤2.5 μm.

Criteria pollutants were positively associated with PD as expected from prior results, though the magnitude of the estimated effect sizes were smaller than those observed for the traffic-related metals and VOCs. For NO2, a standard marker of traffic-related exhaust pollution, 10-year average exposure at the residential addresses was weakly associated with PD (OR = 1.06, 95% CI = 0.92–1.23; Table 2), and a stronger signal was seen at the workplace addresses (OR = 1.27, 95% CI = 1.05–1.54). PM2.5, a multisource mixture that includes traffic-related components, followed a similar pattern such that the 10-year window suggested a weak association at the residential-address (OR = 1.06, 95% CI = 0.95–1.17) and, again, a stronger workplace-address related association was observed (OR = 1.21, 95% CI = 1.03–1.43). In two-pollutant models, the associations for nickel, zinc, benzene, and 1,3-butadiene remained largely unchanged after adjustment for NO2 or PM2.5 (Figure S4); conversely, the NO2 and PM2.5 associations attenuated toward the null when co-adjusted with any of the individual air toxics but especially nickel our strongest brake and tire wear marker.

3.2. Industrial and urban-related pollutants

In contrast to the traffic-related pollutants, lead and chromium, metals with substantial contributions from industrial processes and legacy environmental contamination, showed little evidence of association with PD in these predominantly agricultural counties. Lead estimates hovered near the null across all averaging periods and address types (10-year residential OR = 0.96, 95% CI = 0.83–1.10; 10-year workplace OR = 1.03, 95% CI = 0.93–1.15; Table 2). Chromium showed a similar pattern, with a 10-year residential address OR of 1.06 (95% CI = 0.90–1.24) and workplace-address OR of 1.04 (95% CI = 0.91–1.19), and the estimates were similar across all three averaging periods. Co-adjustment for other pollutants also did not affect the results for lead and chromium more than minimally (Figure S4).

Results were materially unchanged in sensitivity analyses restricting to participants with both residential and workplace addresses available during the 10-year window (Table S4) and after adjustment for a pesticide score based on the PWAS-identified pesticides (Table S5). In a 1-year lagged analysis excluding the year immediately preceding diagnosis, associations for nickel remained robust at both residential and workplace addresses, while VOC associations were attenuated (Table S6).

4. Discussion

4.1. Traffic-related air toxics and Parkinson’s disease risk

In this population-based case-control study from central California, we found that ambient exposure to air toxics with mobile-source contributions – nickel, zinc, benzene, and 1,3-butadiene – was associated with higher risk of PD. Nickel showed the strongest association, particularly at the workplace address, where 10-year per-IQR exposure was associated with nearly three-fold higher odds of PD. These associations were robust to adjustment for pesticide exposure and persisted in co-pollutant models with NO2 and PM2.5, suggesting that specific traffic-related constituents, rather than the ambient mixture broadly, may be most relevant to PD etiology. By examining individual air toxics alongside criteria pollutants within the same study population, our analysis moves beyond the aggregate exposure metrics used in most prior studies and points toward particular components of traffic-related emissions as potential contributors to PD.

4.2. Biological mechanisms of traffic-related air toxics

Inhaled metal-containing particles from brake and tire wear, including nickel and zinc (Fussell et al., 2022; Lopez et al., 2023), can be transported via olfactory receptor neurons to the olfactory bulb through slow axonal transport (Henriksson et al., 1997; Sunderman, 2001), where they accumulate in neural tissue over time (Ijomone, 2021). Once deposited, these metals exert cumulative neurotoxic effects through sustained pro-oxidant activity, disruptions of mitochondrial dysfunction, and promotion of neuronal apoptosis (Anyachor et al., 2022; Genchi et al., 2020), processes consistent with the slow, progressive neurodegeneration characteristic of PD and the stronger associations we observed at the 10-year averaging window. Moreover, because metal emissions from non-exhaust wear are relatively temporally stable, long-term LUR averages capture spatial contrasts with less misclassification than they do for more volatile pollutants (Figure S3). These mechanisms highlight why reductions in tailpipe emissions alone may not sufficiently address neurotoxic exposures, particularly as electric vehicles reduce exhaust emissions but not metal-rich brake and tire wear particles.

In contrast, VOCs such as benzene and 1,3-butadiene are rapidly metabolized to reactive intermediates, benzene oxide and catechol for benzene (Snyder and Hedli, 1996), and epoxybutene and diepoxybutane for 1,3-butadiene (Kirman et al., 2010), with biological half-lives on the order of hours. Their neurotoxic effects may therefore operate through more acute mechanisms, including neuroinflammation, blood-brain barrier disruption, and oxidative damage to dopaminergic pathways (Belluomo et al., 2025; Zheng et al., 2022), which would be better captured by recent exposure windows. Additionally, ambient VOC concentrations exhibit greater temporal variability than metal-containing particulate emissions, owing to fluctuations in traffic patterns, photochemistry, and fuel composition. Averaging VOC exposures over a decade may therefore smooth meaningful variation and introduce non-differential exposure misclassification that attenuates associations, whereas shorter windows may preserve the actual exposure contrast. Collectively, these biological and methodological considerations suggest that metals and VOCs may contribute to PD risk through complementary mechanisms operating on different timescales – cumulative neurodegeneration for metals and more proximate neuroinflammatory or triggering effects for VOCs during the prodromal phase.

4.3. Comparison with prior PM2.5 findings

Previously (Kwon et al., 2024), we reported results for fine particles and PD risk based on a chemical-transport-satellite fusion model beginning in 1981 (Meng et al., 2019) that generated 10-year average exposure measures with an additional 5-year lag and yielded a larger IQR (5.46 μg/m3) and stronger associations with PD (e.g., work address based OR = 1.59, 95% CI = 1.28–1.99). The current LUR-based PM2.5 model starts its predictions only in 1989 (Su et al., 2024) and restricts us to using only unlagged 10-year averages that produced a smaller IQR (1.76 μg/m3) and effect estimate for particles from traffic (e.g., work address based OR = 1.21, 95% CI = 1.03–1.43). Differences in pollution modeling, historical coverage of the model, and exposure contrasts likely explain the variation in PM2.5 results across our two studies. Nevertheless, regardless of which PM2.5 exposure model was used, we observed positive associations with PD that were stronger at workplace addresses, supporting the overall robustness of the PM2.5-PD findings.

4.4. Industrial and urban air toxics

In contrast, lead and chromium showed no evidence for associations with PD, although this may partly reflect limitations of the LUR models in capturing industrial-source variation in our tri-county study area. The lead model was driven largely by measures of proximity to industrial facilities across California (e.g., industry within 50m), and chromium emissions concentrate in small industrial hotspots that may not be well represented in regional LUR surfaces for these three predominantly rural counties (Figure 1). These null findings should therefore be interpreted with caution.

4.5. Strengths and limitations

A key strength of our study is the use of high-resolution LUR models linked to the lifelong residential and occupational address histories over the 10-year exposure window, addressing a limitation of residence-address only exposure models. Our exposure estimates represent modeled ambient concentrations at geocoded addresses rather than personal exposures and do not capture time actually spent at each location. The stronger associations at the workplace address therefore do not reflect occupational exposures (National Research Council, 1997), but likely arise because Central Valley worksites tend to be closer to traffic corridors, distribution hubs, and oil and gas operations (Marjollet et al., 2015), providing greater spatial contrast in traffic-related pollution than residential areas. This may also be one explanation why prior residential-address only based studies, such as the Nurses’ Health Study (Palacios et al., 2014), reported null associations – as limited exposure contrast and shorter averaging periods may have attenuated any signal. Adjustment for PWAS-identified pesticides, predictors of PD in this population, helped reduce confounding from agriculturally derived neurotoxicants and suggests independent contributions of air toxics to PD risk. Two-pollutant models and exposure windows of different length were used to evaluate the sensitivity of the effect estimates to co-pollutant adjustment and to the choice of exposure-averaging period. Finally, primary PD diagnoses were confirmed by movement-disorder specialists, minimizing outcome misclassification and strengthening the reliability of our findings.

However, our study has several limitations. We lacked monitoring data that was species-level PM2.5 (e.g., black carbon) and any air toxics monitoring data to build LUR surfaces before 1989, limiting our ability to evaluate specific PM2.5 components and a broader set of non-tailpipe pollutants. Our LUR models capture spatial variation in ambient concentrations but are not a source apportionment approach: observed associations with air toxics should therefore be interpreted as reflecting ambient mixtures at locations where these toxics covary, rather than as representing a single source. The study population was predominantly composed of white individuals, which may limit the generalizability of the results to more diverse populations. Although we adjusted for individual educational attainment, we did not additionally adjust for neighborhood-level socioeconomic status (SES). However, in this rural agricultural setting the urban pattern of higher pollution exposure in lower-SES neighborhoods is not expected, and cases and controls were sampled from the same Central Valley communities. In a recent analysis from the same study, additional adjustment for nSES did not materially change air-pollution effect estimates (Kwon et al., 2025). Finally, we could not account for additional exposures or environmental stressors that contribute to PD risk and would result in residual confounding if they are highly correlated with the air toxics we modeled.

5. Conclusion

Our findings from this large population-based case-control study in central California strengthen the evidence that air pollution contributes to PD risk, highlighting the underexplored role of specific traffic-related air toxics – particularly nickel and zinc from brake and tire wear, and benzene and 1,3-butadiene from fuel combustion and evaporative emissions. By distinguishing these individual toxicants beyond criteria pollutants, our results provide insight into which components of the traffic-related mixture may drive PD risk. Importantly, because non-exhaust sources such as brake and tire wear will persist even as the vehicle fleet electrifies, the transition to electric vehicles alone will not eliminate these neurotoxic exposures. Given that there is currently no cure for PD, our results underscore the need to reduce environmental exposures and their sources, including through regulatory attention to non-exhaust particulate emissions. Future research should focus on elucidating the biological mechanisms underlying the observed associations and on interventions to reduce individual and community-level exposures. While individuals can reduce personal exposure through everyday choices, such as exercising away from heavy traffic corridors or indoor when ambient air quality is poor, there is little choice regarding the ambient exposures one receives at the workplace location or residence. Our study may help to identify areas where air quality improvements could reduce the risk of neurodegenerative disease and inform policies to mitigate the burden of PD, particularly in highly impacted areas like the Central Valley.

Supplementary Material

Air Toxics and PD supplement

Acknowledgements

We thank all PD patients and their household members as well as all other study participants for their time and support. This work was supported by the California Air Resources Board (CARB; 21RD004).

References

  1. American Lung Association. 2025. State of the Air 2025. https://www.lung.org/getmedia/5d8035e5-4e86-4205-b408-865550860783/State-of-the-Air-2025.pdf (11 March, 2025).
  2. Anyachor CP, Dooka DB, Orish CN, Amadi CN, Bocca B, Ruggieri F, et al. , 2022. Mechanistic considerations and biomarkers level in nickel-induced neurodegenerative diseases: An updated systematic review. IBRO Neurosci Rep. 13, 136–146. 10.1016/j.ibneur.2022.07.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Atterling Brolin K, Schaeffer E, Kuri A, Rumrich IK, Schumacher Schuh AF, Darweesh SKL, et al. , 2025. Environmental Risk Factors for Parkinson’s Disease: A Critical Review and Policy Implications. Mov Disord. 40, 204–221. 10.1002/mds.30067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Belluomo I, Tarazi M, Lao-Kaim NP, Tai YF, Spanel P, Hanna GB. 2025. Detection of Volatile Organic Compounds as an emerging strategy for Parkinson’s disease diagnosis and monitoring. NPJ Parkinsons Dis. 11, 161. 10.1038/s41531-025-00993-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Boogaard H, Patton AP, Atkinson RW, Brook JR, Chang HH, Crouse DL, et al. , 2022. Long-term exposure to traffic-related air pollution and selected health outcomes: A systematic review and meta-analysis. Environ Int. 164, 107262. 10.1016/j.envint.2022.107262. [DOI] [PubMed] [Google Scholar]
  6. Chen M, Gazze L, DiTraglia FJ, Das R, Nriagu J, Erel Y, et al. , 2025. Environmental lead risk in the 21st century. Commun Earth Environ. 6, 776. 10.1038/s43247-025-02735-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cockburn M, Mills P, Zhang X, Zadnick J, Goldberg D, Ritz B. 2011. Prostate cancer and ambient pesticide exposure in agriculturally intensive areas in California. Am J Epidemiol. 173, 1280–8. 10.1093/aje/kwr003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Councell TB, Duckenfield KU, Landa ER, Callender E. 2004. Tire-wear particles as a source of zinc to the environment. Environ Sci Technol. 38, 4206–14. 10.1021/es034631f. [DOI] [PubMed] [Google Scholar]
  9. Degaffe FS, Turner A. 2011. Leaching of zinc from tire wear particles under simulated estuarine conditions. Chemosphere. 85, 738–43. 10.1016/j.chemosphere.2011.06.047. [DOI] [PubMed] [Google Scholar]
  10. Dorsey ER, Elbaz A, Nichols E, Abbasi N, Abd-Allah F, Abdelalim A, et al. , 2018a. Global, regional, and national burden of Parkinson’s disease, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. The Lancet Neurology. 17, 939–953. 10.1016/S1474-4422(18)30295-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Dorsey ER, Sherer T, Okun MS, Bloem BR, Brundin P, Langston JW, et al. , 2018b. The Emerging Evidence of the Parkinson Pandemic. Journal of Parkinson’s Disease. 8, S3–S8. 10.3233/jpd-181474. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fussell JC, Franklin M, Green DC, Gustafsson M, Harrison RM, Hicks W, et al. , 2022. A Review of Road Traffic-Derived Non-Exhaust Particles: Emissions, Physicochemical Characteristics, Health Risks, and Mitigation Measures. Environ Sci Technol. 56, 6813–6835. 10.1021/acs.est.2c01072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Genchi G, Carocci A, Lauria G, Sinicropi MS, Catalano A. 2020. Nickel: Human Health and Environmental Toxicology. Int J Environ Res Public Health. 17, 10.3390/ijerph17030679. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Health Effects Institute. 2025. State of Global Air 2025. https://www.stateofglobalair.org/resources/report/state-global-air-report-2025 (13 March 2025).
  15. Henriksson J, Tallkvist J, Tjalve H. 1997. Uptake of nickel into the brain via olfactory neurons in rats. Toxicol Lett. 91, 153–62. 10.1016/s0378-4274(97)03885-x. [DOI] [PubMed] [Google Scholar]
  16. Holland R, Khan MAH, Matthews JC, Bonifacio S, Walters R, Koria P, et al. , 2022. Investigating the Variation of Benzene and 1,3-Butadiene in the UK during 2000–2020. Int J Environ Res Public Health. 19, 10.3390/ijerph191911904. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hopke PK, Dai Q, Li L, Feng Y. 2020. Global review of recent source apportionments for airborne particulate matter. Sci Total Environ. 740, 140091. 10.1016/j.scitotenv.2020.140091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ijomone OM. 2021. Chapter Eight - Neurotoxicity of nickel. Place. 263–284. [Google Scholar]
  19. Kang GA, Bronstein JM, Masterman DL, Redelings M, Crum JA, Ritz B. 2005. Clinical characteristics in early Parkinson’s disease in a central California population-based study. Mov Disord. 20, 1133–42. 10.1002/mds.20513. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kirman CR, Albertini RJ, Sweeney LM, Gargas ML. 2010. 1,3-Butadiene: I. Review of metabolism and the implications to human health risk assessment. Crit Rev Toxicol. 40 Suppl 1, 1–11. 10.3109/10408444.2010.507181. [DOI] [PubMed] [Google Scholar]
  21. Krzyzanowski B, Mullan AF, Turcano P, Camerucci E, Bower JH, Savica R. 2024. Air Pollution and Parkinson Disease in a Population-Based Study. Jama Network Open. 7, 10.1001/jamanetworkopen.2024.33602. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kwon D, Paul KC, Kusters C, Wu J, Bronstein JM, Lill CM, et al. , 2025. Interaction Between Traffic-Related Air Pollution and Parkinson Disease Polygenic Risk Score. JAMA Network Open. 8, e250854. 10.1001/jamanetworkopen.2025.0854. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Kwon D, Paul KC, Yu Y, Zhang K, Folle AD, Wu J, et al. , 2024. Traffic-related air pollution and Parkinson’s disease in central California. Environmental Research. 240, 117434. 10.1016/j.envres.2023.117434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lopez B, Wang X, Chen LA, Ma T, Mendez-Jimenez D, Cobb LC, et al. , 2023. Metal contents and size distributions of brake and tire wear particles dispersed in the near-road environment. Sci Total Environ. 883, 163561. 10.1016/j.scitotenv.2023.163561. [DOI] [PubMed] [Google Scholar]
  25. McElroy JA, Remington PL, Trentham-Dietz A, Robert SA, Newcomb PA. 2003. Geocoding addresses from a large population-based study: lessons learned. Epidemiology. 14, 399–407. 10.1097/01.EDE.0000073160.79633.c1. [DOI] [PubMed] [Google Scholar]
  26. Meng J, Li C, Martin RV, van Donkelaar A, Hystad P, Brauer M. 2019. Estimated Long-Term (1981–2016) Concentrations of Ambient Fine Particulate Matter across North America from Chemical Transport Modeling, Satellite Remote Sensing, and Ground-Based Measurements. Environ Sci Technol. 53, 5071–5079. 10.1021/acs.est.8b06875. [DOI] [PubMed] [Google Scholar]
  27. National Research Council. 1997. Environmental Epidemiology, Volume 2: Use of the Gray Literature and Other Data in Environmental Epidemiology. Washington, DC. [PubMed] [Google Scholar]
  28. O’Sharkey K, Mitra S, Chow T, Paik SA, Thompson L, Su J, et al. , 2025. Prenatal exposure to criteria air pollution and traffic-related air toxics and risk of autism spectrum disorder: A population-based cohort study of California births (1990–2018). Environ Int. 201, 109562. 10.1016/j.envint.2025.109562. [DOI] [PubMed] [Google Scholar]
  29. Palacios N, Fitzgerald K, Roberts AL, Hart JE, Weisskopf MG, Schwarzschild MA, et al. , 2014. A Prospective Analysis of Airborne Metal Exposures and Risk of Parkinson Disease in the Nurses’ Health Study Cohort. Environmental Health Perspectives. 122, 933–938. 10.1289/ehp.1307218. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Paul KC, Krolewski RC, Lucumi Moreno E, Blank J, Holton KM, Ahfeldt T, et al. , 2023. A pesticide and iPSC dopaminergic neuron screen identifies and classifies Parkinson-relevant pesticides. Nat Commun. 14, 2803. 10.1038/s41467-023-38215-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Raj K, Das AP. 2023. Lead pollution: Impact on environment and human health and approach for a sustainable solution. Environmental Chemistry and Ecotoxicology. 5, 79–85. 10.1016/j.enceco.2023.02.001. [DOI] [Google Scholar]
  32. Shvachiy L, Geraldes V, Outeiro TF. 2023. Uncovering the Molecular Link Between Lead Toxicity and Parkinson’s Disease. Antioxid Redox Signal. 39, 321–335. 10.1089/ars.2022.0076. [DOI] [PubMed] [Google Scholar]
  33. Sikora J, Ouagazzal AM. 2021. Synaptic Zinc: An Emerging Player in Parkinson’s Disease. Int J Mol Sci. 22, 10.3390/ijms22094724. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Snyder R, Hedli CC. 1996. An overview of benzene metabolism. Environ Health Perspect. 104 Suppl 6, 1165–71. 10.1289/ehp.961041165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Su JG, Shahriary E, Sage E, Jacobsen J, Park K, Mohegh A. 2024. Development of over 30-years of high spatiotemporal resolution air pollution models and surfaces for California. Environ Int. 193, 109100. 10.1016/j.envint.2024.109100. [DOI] [PubMed] [Google Scholar]
  36. Sunderman FW. 2001. Nasal Toxicity, Carcinogenicity, and Olfactory Uptake of Metals. Annals of Clinical & Laboratory Science. 31, 3–24. [PubMed] [Google Scholar]
  37. Wang A, Costello S, Cockburn M, Zhang X, Bronstein J, Ritz B. 2011. Parkinson’s disease risk from ambient exposure to pesticides. Eur J Epidemiol. 26, 547–55. 10.1007/s10654-011-9574-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Wise JP Jr., Young JL, Cai J, Cai L. 2022. Current understanding of hexavalent chromium [Cr(VI)] neurotoxicity and new perspectives. Environ Int. 158, 106877. 10.1016/j.envint.2021.106877. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Yu Y, Su J, Mitra S, Gong Y, Li S, Lin Y, et al. , 2025. Air toxics and late-life dementia and cognitive impairment in Mexican Americans. Environ Res. 287, 123105. 10.1016/j.envres.2025.123105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Zheng J, Pang Y, Zhang Y, Hu W, Yang P, Liu Q, et al. , 2022. Indoor VOCs exposure induced Parkinson-like behaviors through autophagy dysfunction and NLRP3 inflammasome-mediated neuroinflammation. Journal of Hazardous Materials. 440, 129818. 10.1016/j.jhazmat.2022.129818. [DOI] [Google Scholar]
  41. Zheng Z, Li H, Zhang Z, Zhai X, Qin H. 2024. Study on the underlying molecular mechanism of benzene-induced nervous system damage in mice based on tandem mass tag (TMT) proteomics. Toxicol Res (Camb). 13, tfae036. 10.1093/toxres/tfae036. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Air Toxics and PD supplement
NIHMS2217038-supplement-Air_Toxics_and_PD_supplement.pdf (1.1MB, pdf)

RESOURCES