Although association between polycyclic aromatic hydrocarbons (PAHs) exposure and mitochondrial DNA copy number (mtDNAcn) was researched by traditional linear model extensively, most of these studies analyzed independent effect of each PAHs metabolite and adjust for the confounding other metabolites concomitantly, without considering others interactions. As a complex organic pollutant, a reasonable statistical method is needed to study toxic effects of PAHs.

Therefore, we aimed to conduct a novel statistical approach, Bayesian Kernel Machine Regression (BKMR), to explore the effect of PAHs exposure on mtDNAcn among coke oven workers. In this cross-sectional study, the concentrations urinary of PAHs metabolites were measured using high performance liquid chromatography mass spectrometry (HPLC-MS). The mtDNAcn was measured using real-time quantitative polymerase chain reaction (RT-PCR) in peripheral blood of 696 Chinese coke oven workers. The relationship of urinary of PAHs metabolites and mtDNAcn were evaluated by BKMR model. And the results showed a significant negative effect of PAHs metabolites on mtDNAcn when PAHs metabolites concentrations were all above 35th percentile compared to the median and the statistically significant negative single-exposure effect of 2-OHNAP and 2-OHPHE on mtDNAcn when all of the other PAHs are fixed at a particular threshold (25th, 50th, 75th percentile). The changes in log 2-OHNAP and 2-OHPHE from the 25th to the 75th percentile when other PAHs metabolites were at the 50th percentile were associated with change in mtDNAcn of −0.082 (−0.021, −0.124) and −0.048 (−0.021, −0.090) respectively. And evidence of a linear effect of urinary 2-OHNAP and 2-OHPHE were found. Finally, our findings suggested that PAHs cumulative exposures and particularly single-exposure of 2-OHNAP and 2-OHPHE might compromise mitochondrial function by decreasing mtDNAcn in Chinese coke oven workers.