Mixture Analyses on Polycyclic Aromatic Hydrocarbons: An Example Using Semen Quality Biomarkers and Bayesian Kernel Machine Regression

Abstract

This study aimed to assess mixture effects of 16 targeted PAHs on male reproductive health by applying a novel grouping approach to the Bayesian kernel machine regression (BKMR) model. Bay region and molecular weight served as two variables to group the PAH mixture for hierarchical variable selection of BKMR. Personal breathing zone air samples were collected to quantify levels of the PAH compounds. Sperm quality was assessed using motility, viability, morphology, while DNA fragmentation served as the biomarker for assessing sperm DNA integrity. BKMR revealed joint effects of the PAH mixture on sperm quality. The bay region grouping was the important component for estimating associations between exposure to the PAH mixture and sperm DNA fragmentation. The low molecular weight grouping played a less important role for assessing the effects of the PAH mixture on sperm motility and viability. An individual PAH compound, benzo(ghi)perylene, was negatively associated with sperm morphology while benzo(b)fluoranthene and benzo(a)anthracene were positively associated with sperm DNA fragmentation. In conclusion, the BKMR model, along with the grouping approach, was suitable to determine the effects of PAH mixtures on sperm quality and DNA integrity.