Occupational exposure and its mechanistic link to allergic asthma and lung function decline; a data-driven approach coupled to mining of adverse outcome pathway signatures

Rob Stierum^1*Manosij Ghosh²Marjolein Meijerink¹Xavier Pinho¹Vivi Schlunssen³Anjoeka Pronk¹Jolanda Van Bilsen¹

• ¹Risk Analysis For Products In Development, Netherlands Organisation for Applied Scientific Research (TNO), Zeist, Netherlands
• ²KU Leuven, Environment and Health, Leuven, Belgium
• ³Aarhus University, Aarhus, Central Denmark Region, Denmark

Within occupational epidemiology, the establishment of associations between chemical exposures and health outcome, in particular of individual chemicals present in the exposome, is difficult. Epidemiological studies are valuable but may be prone to confounders, or lack detailed exposure characterisation. Rodent studies may suffer from interspecies differences in comparison to humans. Here, we explore if a data driven approach can leverage human relevant mechanistic information to inform presumed associations between chemical exposures and two common respiratory disorders: lung function decline (LFD) and allergic asthma (AA). Using public toxicogenomics datasets, we identified Gene Ontology Bioprocesses (GO BPs) enriched in human respiratory cells, exposed in vitro to either diesel ultrafine particles (UFP) or respiratory sensitisers. In addition, for LFD and AA, GO BPs were curated from Molecular Initiating Events (MIEs) and Key Events (KEs) extracted from the Adverse Outcome Pathway (AOP) Wiki, and DisGeNET, a gene-disease database. Considering the commonality in GO BPs, a clear overlap was observed between GO BPs derived from UFP and LFD (a.o. "negative -"/"positive" regulation of cell activation", "positive regulation of ion transport", "stem cell proliferation"). 20 GO BPs were overlapping between sensitisers in combination with AA (e.g. "responses to xenobiotic stimulus", "response to oxidative stress" and "regulation of response to cytokine stimulus"). For AA, sensitiser concentrations used in in vitro were generally higher compared to equivalent concentrations expected in vivo (from PBK modelling). Yet, the overlapping GO BPs discovered for these endpoints were plausible and aided in providing mechanistic insights. Currently, limitations exist in the approach to infer causality (e.g. data availability, coverage of AOPs, in vitro to in vivo dosimetry issues), however it can inform on the identification of chemicals within the occupational exposome and putative mechanistic linkage with occupational diseases. Finally, the annotated MIEs and KEs for LFD and AA may serve as valuable resource for further AOP developments.