Revealing the complex association between environmental toxicants and preserved ratio impaired spirometry: an exploratory study

Preserved ratio impaired spirometry (PRISm), as a novel heterogeneous pulmonary function pattern, is associated with the early stage of chronic obstructive pulmonary disease (COPD). However, whether human exposure to environmental toxicants is associated with PRISm has not been comprehensively studied, particularly with respect to combined exposure to multiple toxicants. The objective of the study was to investigate the effects of single and combined environmental toxicants on PRISm.

We analyzed data from 1,754 adult participants in an American population study. We employed multivariable logistic regression, weighted quantile sum (WQS) regression, quantile g-computation (Qgcomp), Bayesian kernel machine regression (BKMR), and machine learning models to examine associations between 21 environmental toxicants and PRISm. We additionally conducted subgroup analyses stratified by smoking status. We also explored the potential mediating roles of the lipid indicator non-high-density lipoprotein to high-density lipoprotein ratio (NHHR) and lipid-related inflammatory markers.

In the single-exposure model, eight environmental toxicants were found to be positively associated with PRISm, with lead (Pb) exhibiting the strongest association estimate. The results of WQS regression, Qgcomp, and BKMR consistently indicated that mixtures of environmental toxicants showed a stable positive association with PRISm in both the overall population and smokers. Pb and cadmium (Cd) were identified as the most contributory environmental toxicants, respectively. The mediation analysis indicated that NHHR statistically accounted for 12.5% of this relationship, while the mediation proportion of lipid-related inflammatory markers ranged from 8.51% to 11.1%. Gradient boosting decision tree (GBDT) emerged as the top-performing machine-learning model, with the highest area under the curve (AUC) of 0.865. Shapley Additive Explanations (SHAP) interpretability analysis further revealed that Pb was the most influential environmental toxicant.

Our findings provide exploratory evidence of individual and combined associations between environmental toxicant exposure and PRISm in the selected National Health and Nutrition Examination Survey (NHANES) analytic sample.