National Institute for Public Health and the Environment 1 Integrated probabilistic risk assessment Bas Bokkers National Institute for Public Health and the Environment (RIVM) – the Netherlands Deterministic risk assessment -Variability extreme consumer sensitive subpopulations -Uncertainty limited concentration data interspecies extrapolation Worst-case / conservative approach using point values A deterministic risk assessment does not discriminate between variability and uncertainty National Institute for Public Health and the Environment 2 Deterministic risk assessment PoD ADI = AF1 AF2 ….. AFi * National Institute for Public Health and the Environment 3 * * Exposure = consumption * concentration Risk if exposure > ADI or ADI <1 exposure Conclusions deterministic RA Inconclusive: - Exposure is slightly higher than ADI “risks cannot be excluded” Qualitative: - Exposure > ADI risk everyone affected? Remaining question: Quantify the risk: *Percentage of population affected ? Quantify the uncertainty National Institute for Public Health and the Environment 4 Probabilistic risk assessment - Variability extreme consumer sensitive subpopulations - Uncertainty limited concentration data interspecies extrapolation Realistic approach using distributions A probabilistic risk assessment can discriminate between variability and uncertainty National Institute for Public Health and the Environment 5 Integrated probabilistic risk assessment: Evaluates - both variability and uncertainty (but separately) -in both exposure assessment hazard characterization - in a single (integrated) analysis For instance: Combine variability in exposure with variability in sensitivity Combine uncertainty in concentrations with uncertainty in interspecies differences National Institute for Public Health and the Environment 6 Probabilistic risk assessment Individual’s dose that would lead to some predefined effect: PoD of iBMD = AF1 AF2 ….. AFi * * * The same individual’s exposure: of iEXP = consumption concentration * This individual is at risk when his/her iEXP > iBMD or when No information on the individuals…… but variability distributions can inform iBMD and iEXP distributions National Institute for Public Health and the Environment 7 iBMD <1 iEXP Probabilistic risk assessment iBMD <1 An individual is at risk when his/her iEXP > iBMD or when iEXP iBMD distr. = 1 iEXP distr. * Fraction of the population affected National Institute for Public Health and the Environment 8 Probabilistic risk assessment PoD of iBMD = AF1 AF2 ….. AFi * * * of iEXP = consumption concentration * Uncertainty distributions can inform uncertainty in iBMD and iEXP distributions National Institute for Public Health and the Environment 9 PoD distribution BMD distribution PoD distr. iBMD = AF1 AF2 ….. AFi 450 * * 200 250 300 BW Critical effect size (CES) X% decrease in BW 350 400 * 0 500 1000 1500 dose BMD distribution National Institute for Public Health and the Environment 10 Assessment factors PoD distr iBMD = AF1 AF2 ….. AFi * * * Interspecies Subchronic-to-chronic based on historical data (BMD ratios)* Subacute-to-chronic Sensitivity in whole Intraspecies population: Variability 1 Uncertainty about the variability *see e.g. National Institute for Public Health and the Environment 11 Bokkers and Slob tox sci 85 & crit rev toxicol 37 Kramer et al. regul toxicol pharm 23 See van der Voet et al. food chem tox 47 Integrated probabilistic hazard characterization PoD distr. iBMD = AF1 AF2 ….. AFi * * * = * * …… * Variability and uncertainty in these distributions are analyzed separately National Institute for Public Health and the Environment 12 Integrated probabilistic risk assessment iBMD <1 An individual is at risk when his/her iEXP > iBMD or when iEXP iBMD distr. = 1 iEXP distr. * Fraction of the population affected * Uncertainty can be quantified National Institute for Public Health and the Environment 13 Example of integrated prob. RA output Lower Percentile Upper percentile Prob Det (% affected & CI) 0.0001 (0-0.005) effect A no risk effect B risk not excl 0.0001 (0-0.8) effect C risk 0.1 (0-20) effect D risk 8 (5-20) 10 National Institute for Public Health and the Environment 14 iBMD =1 iEXP 100 1000 100 100 75 75 50 50 25 25 cie s In t ra sp e cie s er sp e In t BM D n nc en tra tio Co ns um pt io n 0 Co % contribution to uncertainty Contribution to uncertainty 0 Consumption Concentration BMD Guidance to reduce uncertainty in the RA National Institute for Public Health and the Environment 15 Interspecies Intraspecies Applied in • European projects • Peer reviewed journals • RA advise to Dutch government Limitations • Not implemented yet: approach for carcinogens • More time-consuming (vs lower tier deterministic RA) • Limited no. of uncertainties incorporated Future challenges • Extend approach for carcinogens • Increase acceptance How…..? National Institute for Public Health and the Environment 16 All ingredients are available • Dose-response modeling / BMD techniques are available • Empirical AF distributions are available (excl. intraspecies AF) • Probabilistic exposure assessment techniques are available • Integration techniques are available Limited tox or exposure data? Larger uncertainty Incorporated in probabilistic RA And…….. National Institute for Public Health and the Environment 17 Benefits of (integrated) probabilistic RA • Quantification of - Fraction of the population affected - Uncertainty • Risks can be compared - between effects - between substances • Probabilistic approach provides more insight in risk Targeted risk management actions or further research National Institute for Public Health and the Environment 18 Thank you for your attention National Institute for Public Health and the Environment 19 Further reading • Bokkers, B. et al (2009). The practicability of the integrated probabilistic risk assessment (IPRA) approach for substances in food. RIVM report 320121001/2009, Bilthoven, the Netherlands. http://www.rivm.nl/bibliotheek/rapporten/320121001.pdf • Bosgra, S. et al (2009). An integrated probabilistic framework for cumulative risk assessment of common mechanism chemicals in food: an example with organophosphorus pesticides. Regul Toxicol Pharmacol 54, 124-33. • Müller, A.K. et al (2009). Probabilistic cumulative risk assessment of antiandrogenic pesticides in food. Food Chem Toxicol 47, 2951-62. • van der Voet, H. and Slob, W. (2007). Integration of probabilistic exposure assessment and probabilistic hazard characterization. Risk Anal 27, 351-71. • Benchmark dose software: www.proast.nl • EFSA (2009) Guidance of the Scientific Committee: use of the benchmark dose approach in risk assessment. The EFSA Journal 1150, 1-72 http://www.efsa.europa.eu/en/scdocs/scdoc/1150.htm bas.bokkers@rivm.nl National Institute for Public Health and the Environment 20