Moderator: John Greally
Rapporteur: Madeleine Laffont
This question is linked with how you design the system that you want to study in the first place: what organism, what exposure etc. Discussions on this topic covered dose-response issues, developmental timing of how you do the exposure and how you sample the cells. The following points were agreed as necessary when designing epigenetic studies:
Multidisciplinary / consortium projects, which include a cross section of disciplines in the design phase as well as the implementation phase (experts from the fields of regulatory science to ensure the project is relevant to regulatory toxicology; developmental biology, epidemiology, computational and bioinformatics, toxicology, epigenetics, biochemistry, medicine, pharmacology etc). For example, developmental biologists will be needed since the cells that mediate the particular phenotype in question may disappear during development as a result of the toxicant exposure and not be available for study. The epigenome and transcriptome of these developmental cells should be assessed by performing testing at the appropriate developmental time point.
Detailed histology and full understanding of the cell sub-types present are needed in order to understand the molecular assays.
Reproducibility: the system set up in terms of exposure and outcome will work equally well in multiple different labs because we are studying variability – and if the variability is dependent upon the lab, then interpretation of results is impossible.
The experimental organism choice: consider organism specific issues. For example, zebrafish do not have a uterus; in rat some of the reference genomes needed for particular strains of rat is very poor – therefore a rat genome sequencing approach to upgrade the genomes of some of the rat strains consistently may be required).
Two phased approach to Study design: the first phase will be expensive – but as soon as it is informative it can be transitioned into cost-effective second-phase approaches:
Phase 1: Experimental animal studies will include: DNA methylation, hydroxymethylation, chromatin analysis (helps with interpretability of other data), transcriptional analysis, genotype of the experimental animal to create interpretable molecular data. Additional expenses, such as cell phenotyping (flow cytometry) etc, will also be needed to adequately power the study, ensure it is useful, and provide the molecular markers needed for the overall phenotype
Phase 2: more cost effective approaches over scale, based on phase 1 findings.
Model compounds: the compounds mentioned by Group 1, but also suggested DES – because, although less relevant to human health today, it affects cells that persist and ultimately give rise to the adverse phenotype of cancer in humans. This has the advantage over some of the other suggested model compounds where the adverse outcome is related to phenotype and therefore the cells ultimately disappear.
Regulatory perspectives: the epigenetic study must have added value over and above other accepted tests in terms of insights or predictive capacity, and the model system must be applicable to humans.
Consistent and standardised data management: consistent analysis, and terminology necessary for reproducibility of results. This has impacts on the choices of software and parameters used and metadata are collected systematically (transparency in experiment design).
Moderator: John Greally