Three PM2.5 Chemicals Emerge as Prime Suspects in Age-Related Cognitive Decline

Fine particulate matter smaller than 2.5 micrometers across, known as PM2.5, has long been linked to cognitive decline, but the microscopic cloud of industrial chemicals hitching a ride on those particles has remained largely unexamined. A new cross-sectional study from China now takes one of the most detailed looks yet at that chemical cargo, measuring 100 distinct PM2.5-bound compounds and testing which of them track most closely with impaired cognition in middle-aged and older adults. Using a battery of statistical models, the researchers narrowed a sprawling field of candidates down to three chemicals of particular concern: 2-methyl-benzothiazole, isopropyl myristate, and 2,6-di-tert-butyl-4-(dimethylamino-methyl)phenol, better known by its trade name Ethanox-703.

The analysis drew on baseline data from the Sub-Clinical Outcomes of Polluted Air in China (SCOPA-China) study, a multicenter prospective cohort. For this cross-sectional slice, the team enrolled 373 participants aged 40 to 89 years living in seven Chinese cities: Shijiazhuang, Harbin, Wuhan, Wuxi, Jinan, Xi’an, and Chengdu. Each participant had lived at their current address for at least two years, and people with hearing or language impairments were excluded so that cognitive testing could be administered reliably. Written informed consent was obtained from all participants, and the study received ethical approval from the National Institute of Environmental Health at the Chinese Center for Disease Control and Prevention.

Measuring exposure was a central technical challenge. Rather than relying on modeled estimates, the researchers installed PM2.5 samplers on rooftops in the residential communities where participants lived, keeping sites at least five kilometers from major industrial sources. Each sample was collected over roughly 24 hours across three to five consecutive days, captured on quartz filters at a flow rate of 100 liters per minute. Chemical analysis then employed ultra-performance liquid chromatography coupled with triple quadrupole mass spectrometry, or gas chromatography with single quadrupole mass spectrometry, to quantify the particle-bound chemistry. In total, 100 chemical constituents with detection rates above 80 percent were identified and sorted into 14 categories, including phthalate esters, fatty acid ester plasticizers, benzothiazoles and benzotriazoles, benzophenones, synthetic antioxidants, bisphenols, organophosphate esters, and parabens.

Cognitive function was assessed with the validated Chinese version of the Mini-Mental State Examination, a 30-item instrument covering orientation, recall, attention and calculation, language, and visuospatial coordination. Cognitive impairment was defined as an MMSE score below 27, a threshold drawn from established clinical criteria for Chinese populations. Within the study population, 65 individuals met that definition, while the overall average MMSE score was 28.00. The cohort was nearly evenly split by sex, with 180 men and 193 women, and the mean age was 63.86 years. During the survey period, the daily average PM2.5 concentration measured in the communities was 103.03 micrograms per cubic meter, and the one-year moving average stood at 62.50 micrograms per cubic meter.

The statistical architecture of the study was deliberately layered. Generalized estimating equations served as the primary single-pollutant model, with sampling site as the clustering unit and small-sample bias-corrected robust standard errors to account for the limited number of clusters. To avoid overfitting, the team combined LASSO regression with directed acyclic graphs to select a minimal confounder set, which settled on age, sex, education, income, body mass index, marital status, and occupation. Exposure increments were expressed as a tenfold, or log10-transformed, increase in each chemical’s concentration. On top of the primary model, the researchers ran generalized linear mixed-effects models with nested random intercepts, ordinary generalized linear models, and a variant of the primary model using the continuous MMSE score rather than the dichotomized impairment outcome. False-discovery-rate correction was applied across all single-pollutant tests.

Thirteen chemicals emerged with suggestive, FDR-corrected associations with cognitive impairment in the primary single-pollutant models, and all thirteen odds ratios exceeded one, indicating that higher exposure corresponded to higher odds of impairment. The list included dibenzyl phosphate, tert-butylphenyl diphenyl phosphate, 2-ethylhexyl sebacate, isopropyl myristate, 4,4′-bis(dimethylamino)benzophenone, tris(4-tert-butylphenyl) phosphate, 2-methylbenzothiazole, Ethanox-703, 5-chloro-1H-benzotriazole, glycerol monooleate, dipropylene glycol dibenzoate, isopropyl palmitate, and ethylhexyl methoxycinnamate. Among the seven chemicals that held up across multiple model specifications, tris(4-tert-butylphenyl) phosphate showed the largest odds ratio at 1.70 (95 percent confidence interval 1.27 to 2.29), followed by 2-ethylhexyl sebacate at 1.66 and 2-methylbenzothiazole at 1.57.

Because single-pollutant models struggle with the correlated mixtures that characterize real air pollution, the team added two mixture-exposure approaches: weighted quantile sum regression and quantile g-computation. Each model was fitted using all chemicals within a predefined group. In the weighted quantile sum analysis, four groups showed significant positive associations with cognitive impairment after FDR correction: fatty acid ester plasticizers (odds ratio 1.47), adipate esters (1.30), synthetic antioxidants (1.54), and the benzothiazole and benzotriazole group (1.44). Quantile g-computation, which allows chemicals to contribute in either direction, confirmed significant positive associations for the fatty acid ester plasticizers, mono-phthalate esters, synthetic antioxidants, alkyl organophosphate esters, and benzothiazole and benzotriazole groups. Three groups overlapped across both mixture methods, and within them, isopropyl myristate carried a weight of 0.94 in the weighted quantile sum model, 2-methylbenzothiazole carried 0.57, and Ethanox-703 carried 0.54, with Ethanox-703 rising to 0.72 in the quantile g-computation output.

The intersection of these analyses singled out the three headline compounds. 2-Methyl-benzothiazole belongs to the benzothiazoles, high-production-volume chemicals used chiefly as vulcanization accelerators in rubber manufacturing and also as fungicides, corrosion inhibitors, and ultraviolet stabilizers. Isopropyl myristate is an ester of isopropyl alcohol and myristic acid, widely used as an emollient and solvent in creams, lotions, and oils. Ethanox-703 is a synthetic phenolic antioxidant employed as an oxidation inhibitor in elastomers, polyolefin plastics, resins, adhesives, petroleum oils, and waxes. Notably, no directly relevant published research on the effects of isopropyl myristate itself on cognition, memory, or neurotoxicity could be retrieved by the authors, underscoring how little is known about these ubiquitous compounds.

To contextualize their findings, the authors turned to indirect toxicological evidence. For the benzothiazoles, prior work has shown that benzothiazole can alter GABA type A receptors expressed in frog oocytes, potentiating inhibitory signaling, and related research has implicated excessive GABA production during amyloid-beta plaque clearance in neuronal suppression and memory impairment. For isopropyl myristate, the reasoning runs through its precursor, isopropyl alcohol: toxicological experiments in rodents have shown that acute isopropanol administration lowers dopamine in the hypothalamus, pons medulla, and cerebral cortex, reduces noradrenaline across examined brain areas, and decreases serotonin in the striatum and hippocampus in a dose-dependent manner. All three neurotransmitter systems are implicated in memory and cognitive function, and noradrenergic degeneration in particular is a recognized feature of Alzheimer’s disease pathology. For Ethanox-703, the authors pointed to studies of structurally related synthetic phenolic antioxidants such as BHT, BHA, and 2,4-di-tert-butylphenol, which have disturbed serotonergic and dopaminergic signaling in zebrafish development and suppressed the proliferation of normal human astrocytes through cell cycle arrest and endoplasmic reticulum stress pathways.

The authors are careful about what these results can and cannot show. As a cross-sectional design, the study reveals associations, not causal relationships, and cannot rule out reverse causation. Only 65 of the 373 participants had cognitive impairment, limiting statistical power for subgroup analyses. Community-based fixed-site sampling without biomarker validation risks exposure misclassification, and single-time-point sampling cannot capture long-term cumulative exposure. The MMSE alone does not probe specific cognitive domains such as executive function or episodic memory, and the neurotoxic mechanisms invoked in the discussion are inferred from animal and cellular studies rather than validated in humans. The mixed-effects sensitivity models also showed near-zero random-effect variances and frequent singular-fit warnings, leading the authors to treat those outputs as exploratory. Still, the convergent signals across single-pollutant and mixture models, and the negative associations between the three candidate chemicals and continuous MMSE scores in the sensitivity analysis, mark 2-methyl-benzothiazole, isopropyl myristate, and Ethanox-703 as priority targets for the mechanistic and prospective research that will be needed to determine whether these airborne chemicals genuinely threaten the aging brain.

Subject of Research: Associations between PM2.5-bound emerging chemical constituents and cognitive impairment in middle-aged and older adults in China

Article Title: Multi-model assessment reveals drivers among PM2.5-bound emerging chemical constituents impacting cognitive health of middle-aged and older adults

Article References: Shen, Y., Wang, L., Du, P., Chen, X., Liu, Y., & Li, T. (2026). Multi-model assessment reveals drivers among PM2.5-bound emerging chemical constituents impacting cognitive health of middle-aged and older adults. Frontiers in Environmental Science, 14, Article 1977587. https://doi.org/10.3389/fenvs.2026.1977587

Image Credits: AI Generated

DOI: 10.3389/fenvs.2026.1977587

Keywords: PM2.5, air pollution, cognitive impairment, MMSE, benzothiazoles, synthetic antioxidants, plasticizers, chemical exposome, cross-sectional study, China, aging, mixture models