Hristo Aladjov, Brussels, 2 Sep. 2018 & Quantitative AOPs Hristo Aladjov, Brussels, 2 Sep. 2018
Effectopedia – Intersection of frameworks ADME AOP Effectopedia – Intersection of frameworks q MIE Aromatase Inhibition KE1 Reduced E2 synthesis AO Population Reduction … Stressor Fadrozole internal Fadrozole external Experimental In-Vitro f(x) In-Vivo In-Vitro f(x) In-Silico Model f(x) Modelling
ADME AOP ADME & qAOP q MIE Aromatase Inhibition KE1 Reduced E2 synthesis AO Population Reduction … Stressor Fadrozole internal Fadrozole external time dose Stressor (external) time dose Stressor (at target) time response MIE time response KE1 time response AO The input of the qAOP model is the dose-time data for the stressor measured at the target site Use toxicokinetics to extend the predictive power of qAOP (so they can be used as alternative to in-vivo animal models)
Experimental evidences - qAOPs time dose Stressor (at target) time response KE1 MIE Aromatase Inhibition KE1 Reduced E2 synthesis AO Population Reduction … time dose Stressor (at target) time response MIE MIE Aromatase Inhibition KE1 Reduced E2 synthesis In-Vitro f(x) In-Vivo In-Vitro f(x) time Assay response In-Vitro1 time dose Stressor (Media 1) f(x) In-Vitro 1 time Assay response In-Vitro 2 time dose Stressor (Media 2) f(x) In-Vitro 2
Experimental evidences - qAOPs KE need to be measurable we know how to interpret a measurement we know how define the test response mapping. Test response mapping transforms the output of the assay into in-vivo relevant data, comparable across KE In case of chemical stressors ADME is important basis for comparing the assay and in-vivo model thus helps define the test response mapping. MIE Aromatase Inhibition KE1 Reduced E2 synthesis AO Population Reduction … MIE Aromatase Inhibition KE1 Reduced E2 synthesis In-Vitro f(x) In-Vivo In-Vitro f(x) f(x) In-Vitro 1 f(x) In-Vitro 2
Experimental design and modelling Creating an qAOP is iterative process Quantify the AOP using the existing data Create initial models Identify data gaps and uncertainties Measure and publish the data which will be useful for parametrization of the models. MIE Aromatase Inhibition KE1 Reduced E2 synthesis AO Population Reduction … In-Vitro f(x) In-Vivo In-Vitro f(x) In-Silico Model f(x)
qAOP networks – the functional unit of predictions Chemical 1 MIE1 KE KE KE AO1 Chemical 2 MIE2 KE KE AO2 Chemical 3 MIE3 KEX KE AO3 Chemical 4 MIE4 KEY KE AO4 Chemical 5 MIE5 KE KE KE AO5 Analysis of the most sensitive AO Single MIE can trigger multiple AOs Alternative test prioritisation Toxicity of mixtures Chemical2 Chemical4 Chemical5 concentration
In-silico models in Effectopedia Several types of models - dose response, response response, ADME, … Models use input, output and model parameter conventions to integrate with Effectopedia No restrictions on the methods used – curve fitting, statistical, system biology models, deep learning … Currently supported R, MATLAB, Java, more to come Single model can cover more than one KE Feedbacks / feedforward should be handled internally in the model covering all involved KE MIE Aromatase Inhibition KE1 Reduced E2 synthesis KE2 Reduced E2 plasma concentrations HPG axis model