Work

Urinary polycyclic aromatic hydrocarbons and depressive/anxiety symptoms in older adults: an exploratory machine learning analysis

Year: 2026

Type: article

Abstract: Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous neurotoxic pollutants, yet evidence linking PAHs to older adults’ depressive/anxiety symptoms remains scarce. We explored single/mixed urinary PAH-metabolite associations as well as sex- and body-mass index (BMI)-related effect modification among 413 Shenzhen adults aged ≥ 60 years. Symptoms were assessed using the Geriatric Depression Scale (GDS) and Hamilton Anxiety Scale (HAMA). Complementary machine-learning analyses with XGBoost and SHapley Additive exPlanations explored key hydroxylated polycyclic aromatic hydrocarbon (OH-PAH) metabolites to cross-check conventional regression findings. Logistic regression showed that urinary 1-hydroxypyrene (1-OHPyr) and summed hydroxylated phenanthrene metabolites (ΣOH-PHEs) were associated with higher odds of prevalent anxiety symptoms. Sex-stratified analyses linked 3-hydroxyphenanthrene (3-OHPhe) to prevalent anxiety symptoms and 1-OHPyr to prevalent depressive symptoms in females, and stronger 1-OHPyr–anxiety associations were observed in non-overweight participants. Linear regression revealed a significant 1-OHPyr–HAMA association, whereas only a marginal non-significant trend was seen for GDS scores. Bayesian kernel machine regression indicated higher odds of prevalent anxiety symptoms at high mixture exposure, whereas quantile g-computation showed no significant association. The machine-learning analysis flagged 1-OHPyr as the top predictor for anxiety, with suggestive signals for depressive symptoms. Associations between phenanthrene-derived OH-PAHs and anxiety were subject to BMI modification. However, after false discovery rate (FDR) correction, these associations did not remain statistically significant (smallest PFDR = 0.494); the findings should be regarded as hypothesis-generating. This simultaneous assessment of PAH-related mood-symptom associations in older adults highlights the need for prospective cohorts to verify findings and clarify underlying mechanisms.

Source: Journal of Environmental Exposure Assessment

DOI: 10.20517/jeea.2026.47