Benchmark dose modelling (BMDL)

Traditionally, the hazard and risk assessor has attempted to identify the critical no-observed-adverse-effect level (NOAEL) from the available dataset from which an acceptable or tolerable human dose (e.g. ADI/TDI) can be extrapolated (for substances which are not genotoxic carcinogens). Where adverse effects were observed at all test doses, a lowest-observed-adverse-effect level (LOAEL) is sometimes used as the Reference Point (RP)/point-of-departure (POD) for risk assessment purposes. Both approaches have limitations (particularly extrapolating from a LOAEL), notably that the critical effect level (NOAEL/LOAEL) is restricted to one of the predefined dose levels used in the study. In addition, these approaches fail to take into account directly the shape of the dose-response curve, thus all available information is not utilised. Since studies with low power (e.g. small group sizes) and/or insensitive methods are able to detect only relatively large effects, these tend to result in higher NOAELs than would perhaps have been reported in better quality studies.

However, the Benchmark Dose (BMD) approach aims at estimating the dose that corresponds to a low, but measurable change in response (BMDR; e.g. 5% increase in relative kidney weight (continuous data) or a 10% increase in the incidence of liver tumours (quantal data) from the modelled background response). The choice of 10 or 5% for the BMDR is dependent on toxic effect and conservatism of evaluator and is calculated by fitting mathematical models to the dose-response data. A number of models are run, and a lower 95% confidence limit calculated (BMDL; to take account of the quality of the study). From those models considered compatible with the data (i.e. that pass the goodness-of-fit test) the lowest BMDL is typically used as the RP/POD. At least 2 test groups and a control, together with a dose-response relationship, are needed in order to use this approach. The two main models are the US EPA BMD software and The National Institute for Public Health and the Environment of the Netherlands (RIVM) software for dose-response modelling and benchmark dose analysis (PROAST).

Overall, the BMD approach is considered a more sophisticated and powerful approach for determining a RP/POD as it makes use of a greater number of characteristics of the study under examination. Critically, the BMD approach can be used for genotoxic carcinogens. Indeed, an EFSA Scientific Committee recommended in 2011 that EFSA Scientific Panels and Units adopt the BMD approach for the risk assessment of chemicals in food. We can run the models, provide the outputs and help interpret the results.


Our team of experienced toxicologists has used BMD analysis in hazard and risk assessments for a UK government agency and in assessments for industry across multiple sectors



Recent work in this field

More about us How we can help Our recent work