Where are we going with chemical risk assessment? The challenges for the future

Transcription

1 Training Event E Wastewater treatment by advanced technologies and risk assessment framework 1 st ANSWER Workshop Risk prognosis of environmental and public health aspects of antibiotics and antibiotic-resistant bacteria and antibiotic resistance genes (A&ARB&ARGs) September 4-6, 2017, Fisciano (SA) Italy Where are we going with chemical risk assessment? The challenges for the future Emanuela Testai Istituto Superiore di Sanità Environment and Health Department Rome, Italy emanuela.testai@iss.it 1

2 Current safety testing methods Phys-chem properties Toxicological profile vs Exposure Risk assessment The present RA paradigm generally focuses on hazard identification and characterisation as first steps. There is a demand for changing the basis of RA, giving more focus on 1) modes of action (mechanistic approach) 2) a progressive reduction of tests using laboratory animals 3) exposure driven process Towards the Tox21 and the EU SC document on New challenges for RA (2013)

3 ANSWER Workshop 2017 Exposure: Advances in exposure assessment 1. Determination of the internal dose plus a shift from single to individual/groups of chemicals 2. Exposure data increasingly based on human surveillance data (biomonitoring) using biomarkers of exposure 3. Exposome approach: Embracement of the lifetime exposure of a human to chemicals from conception to death: how these exposures relate to the development of disease? Hazard: Reduction in animal use 1. Application of modelling and Non testing methods (TTC or Read across) and the TK-first approach No or limited testing for hazard identification/ characterization in the absence of absorption 2. Development of in vitro preparations maintaining in vivo characteristics over long periods, identification of MoA and key events (AOP) and assessment of in vitro biokinetics 3

4 Blood Concentration ANSWER Workshop 2017 External Exposure Pharmaco/ toxico-kinetics ADME X X* absorption distribution metabolism escretion i.v. Internal Exposure oral Metabolite Time (hrs) The processes integrating ADME, determining the internal dose following external exposure is usually referred to as toxicokinetics (TK) Interaction between X o X* and the target EFFECT Pharmaco/ toxico-dynamics

5

6 The time dependent concentration of a certain pollutant in human tissues can be predicted using physiologically-based pharmacokinetic (PBPK) models. These models consider the human body as a set of well stirred compartments linked by the blood flow. Physiological processes are represented by a set of ordinary differential equations describing the ADME processes of a specific chemical. The final result is a model that simulates the time course distribution of a substance in the human body, which helps to quantify the relationship between measures of external exposure and internal dose 6

7 CSA Ibuprofen PBPK models have the potential for extrapolation from observed kinetic data. The expansion of application of PBPK-models also to in vitro conditions will result in more precise quantification of tolerable exposures and extrapolation from in vitro to in vivo.

8 ecvam%20toxicokinetics%20strategy.pdf Published July 2015 From the ABSTRACT: Information on human toxicokinetics plays an important role in the safety assessment of chemicals, even though there are few data requirements in the EU regulatory framework 8

9 P F O A Female rat t1/2 PFOA = 3 h Male rat t1/2 = 5 d Mouse t1/2 = d Reprotox study on female rats are poorly representative for human extrapolation Human t1/2 = 3.8 ys (average); ys (range). P F O S Male Rat t1/2 PFOS = 43 d Cynomolgus t1/2 = 2.3 ys Human t1/2 =5.4 ys Comparison among studies in different species should be carried out by using the on the basis of the internal dose rather than on the external one.

10 The time dependent concentration of PFOS and PFOA in human tissues can be predicted using PBPK models. The PBPK model was developed by considering the main target tissues of toxicological relevance for PFOA and PFOS Trends in the simulation results indicate that the urinary PFAS resorption-based PBPK model seems to be a reliable approach to explain the relatively longer half-life of PFOA and PFOS in human plasma. How blood levels are representative for tissue levels? The model had been successfully validated by using experimental data in human blood, but good validation results were not achieved for other human tissues (knowledge is very limited; uncertainty and variability) Fabrega et al., Toxicology Letters 230 (2014)

11 An association between PFOA and PFOS serum levels and delayed age at menarche was reported in a cross-sectional study of adolescents. Growth dilution and the new route of excretion (menstruation) could account for some of the reported association. A Monte Carlo (MC) physiologicallybased pharmacokinetic (PBPK) model of PFAS was developed to simulate plasma PFAS levels in a hypothetical female population aged 2 to 20 years old, incorporating realistic distributions of physiological parameters as well as timing of growth spurts and menarche to assess how much of the apparent epidemiological association during puberty can be explained by pharmacokinetic variability. 11 Wu et al, Environment International 82 (2015) 61 68

12 Parameters in the simulated subjects were comparable to those reported in the epidemiologic study. Individual variations in PFAS kinetics associated with rapid growth around the onset of menstruation may contribute to the reported relationship between serum PFAS levels and age at menarche. The reported relationship between PFAS and age at menarche appears to be at least partly explained by pharmacokinetics rather than a toxic effect of these substances. Wu et al, Environment International 82 (2015)

13 Impact of individual s genotypic state on Exposure-Response Continuum EXPOSURE DOSE Response Ambient Monitoring In air, soil, water, diet Biological monitoring In blood, urine Health Surveillance External Exposure Internal Exposure Biochemical Effect Biological Effect Health Impairment, Illness Chemicals Toxins, Metabolites Protein adducts, DNA adducts Cytogenetic, Immunological parameters Genetic Susceptibility significance for risk assessment (Angerer et al., Int J Hyg Envir Heal 2007)

14 Genetic and Phenotipic differences

15

16 Gene-environment interaction Multifactorial diseases result from the interaction of individual susceptibility genetic or acquired factors and exposure to modifiable environmental factors. No Exposure Resistant Genotype Susceptible Genotype Background Risk Level (low) Exposure Resistant Genotype Susceptible Genotype Increased background Risk associated to exposure (ER) n-fold Increased ER

17 Gene-environment interaction Studies at the molecular level in humans suggest that there is an individual variability in genetic parameters, consistent with different susceptibility to disease. An individual may be at high risk due to genetic predisposition or susceptibility. For genetically predisposed individuals risk is strongly dependent on exposure.

18 In Silico non testing methods 1. Read Across principle The toxicological profile of chemical A is known and data are available No or scant data are available for chemical B When it can be demonstrated that A and B are structurally and toxicologically related, it is possible to use data on hazard identification available on A to evaluate hazard caused by B. Obviously for RA the exposure scenarios of B should be considered (which could be different from the A one) The structural analogy should be supported by 1) an in silico analysis (SAR Structure activity relationship) or 2) bridging studies showing a similar toxic behaviour

19 ECHA: Practical Guidance Read-across predicting unknown properties of one chemical from known properties of similar chemicals is a scientific method for filling data gaps on the effects of chemicals. The aim of the Read-Across Assessment Framework (RAAF) is to provide a structured approach to the scientific evaluation of read-across justifications made by registrants in their dossiers. OECD GUIDANCE ON GROUPING OF CHEMICALS, SECOND EDITION Series on Testing & Assessment No. 194 (2014) jm/mono(2014)4&doclanguage=en 19

20 Quaternary ammonium compound Bardap 26 Read across from data of the structurally related QUAT Didecyldimethyl-ammonium chloride (DDAC) is requested for metabolism, teratogenicity/ reproduction, chronic toxicity carcinogenicity, bioaccumulation and chronic ecotoxicity for the active substance Bardap 26. The read across is supported by a set of bridging studies for DDAC demonstrating the similarity in physico-chemical and toxicological properties of these quaternary substances. 20

21 Physico-chemical properties Physico-chemical prop. DDAC Bardap 26 Physical state (at ntp) Light-coloured solid Yellow liquid Melting temperature Melted at C followed by decomposition at ca 280 C. <-50 C. No a melting point or a freezing point down to 50 C. Boiling temperature Decomposition at ca 280 C C Relative density at 20 C at 20 C Vapour pressure 5.9 x 10-6 Pa, 20 C 1.8 x 10-6 Pa, 20 C Henry s Law constant 4.27E-09 Pa m 3 /mol H monomer = 3.03E-11 Pa.m 3 /mol Partition coefficient Not determined as substance is ionic and surface active (~ 1) Not determined as the substance is ionic and surface active (~ 1) Water solubility 500 g/l (20 C ph ca ) Completely miscible with water (> 500 g/l) Dissociation constant Not applicable, the substance is irreversibly ionised. Not applicable, the substance is irreversibly ionised. Surface tension 27.0 mn/m at 20 C (1g/l) 30.5 mn/m at 20 C (1g/l) Solubility in ethanol > 250 g/l at 20 C > 250 g/l at 20 C Solubility in octanol > 250 g/l at 20 C >250 g/l at 20 C Flammability Not highly flammable Not highly flammable Self ignition temperature ca. 195 C > 400 C Explosive properties Non explosive Non explosive Oxidising properties Non oxidising Non oxidising Reactivity towards container Non-reactive Twinning ProjectN. to metals EE05-IB-TWP-ESCand plastics Non-reactive to metals and plastics Tallin April materials

22 The acute hazardous properties of Bardap and DDAC mainly relate to the local effects of the reactive QUATs and are characterized by severe irritation and primary tissue damage by corrosion at the site of application. Other effects are considered to be secondary to local ones. The subchronic toxicity endpoints and health based values (HBV) are in a similar range for Bardap and DDAC. Both compounds were negative in the mutagenicity test battery DDAC has comparable TK behaviour with another structurally related compound with the same MoA, namely ADBAC, as well as HBV for developmental and chronic toxicity and showed no effects in 2-generation and carcinogenicity studies. 22

23 Bridging studies Endpoint DDAC Bardap 26 Acute toxicity LD50 oral rat LD50 dermal rabbit 238 mg/kg >2000 mg/kg 788 mg/kg read across Skin irritation rabbit Eye irritation rabbit corrosive corrosive corrosive read across Sensitization (Buehler) (M+K) not sensitizing not sensitizing not sensitizing Subchronic tox NOAEL 90 day oral rats NOAEL 90 day oral mice NOAEL 8 weeks oral dogs NOAEL 90 day dermal rats 61 mg/kg/d 107 mg/kg/d 30 mg/kg/d 12 mg/kg/d 90 mg/kg/d Mutagenicity Ames Mouse lymphoma cells Chromosome aberration negative negative negative negative negative negative 23

24 Endpoint DDAC ADBAC Bardap 26 Developmental toxicity Rats, oral: NOAEL maternal toxicity NOAEL teratogenicity Rabbits, oral: NOAEL maternal toxicity NOAEL teratogenicity 2-Generations, rats NOAEL parental NOAEL F1 NOAEL F2 Chronic toxicity 104 weeks, rats NOAEL 52 weeks, dog NOAEL Carcinogenicity 104 weeks combined, rats 78 weeks, mice ADME, rats Tallin April mg/kg/d >20 mg/kg/d 10 mg/kg/d >20 mg/kg/d no effects 750 ppm 750 ppm 750 ppm 37 mg/kg/d 10 mg/kg/d no effects no effects <2.5% urine 89-99% faeces <1% in tissues 10 mg/kg/d >100 mg/kg/d 3 mg/kg/d >9 mg/kg/d no effects 1000 ppm 1000 ppm 1000 ppm 44 mg/kg/d 13 mg/kg/d no effects no effects 5-8% urine 87-99% faeces read across read across read across read across read across Twinning ProjectN. EE05-IB-TWP-ESC- 09 <1%in tissues 24

25 Based on The bridging studies results The structural similarity The similar MoA and the results showing similarity between other QUATs (e.g. the similar metabolism pattern of DDAC and ADBAC) it can reasonably be assumed that also for Bardap 26 similar results would be found as they have similar physico-chemical properties and similar chemical structure The read across for the above mentioned toxicological end-points from DDAC data to Bardap 26 is acceptable. 25

26 2. In silico methods In silico is an expression used to mean "performed on computer or via computer simulation." The phrase was coined in 1989 as an allusion to the Latin phrases in vivo, in vitro. They include expert system as QSAR (quantitative structure-activity relationships) models, that is mathematical models that correlates a quantitative measure of chemical structure to either a physical property or a biological effect (e.g., toxic outcome). The term quantitative in QSAR refers to the nature of the parameters (also called descriptors) used to make the prediction. A molecular descriptor provides a means of representing molecular structures in a numerical form. ANSWER Workshop 2017 An assessment of QSAR model validity should be performed by reference to the internationally agreed principles for the validation of QSARs. An important issue in model validation is the definition of its applicability domain. The applicability domain of a QSAR is the physico-chemical, structural or biological space, knowledge or information on which the training set of the model has been developed, and for which it is applicable to make predictions for new compounds. Therefore QSAR models are associated with limitations as 26 to what they can reliably be used

27 3. Threshold of Toxicological Concern (TTC) The TTC approach is a science-based pragmatic tool for screening and prioritizing chemicals for their safety assessment when hazard data are incomplete and human exposure can be estimated. It can be used to evaluate chemicals for which toxicological data are not available (or to prioritise among a number of chemicals those for which there is a need to produce data) It represent a generic toxicological alert threshold giving possible concern, applicable to all chemicals (excluding some specific categories), below which the probability to have significan health risk is very low. The approach was initially developed to evaluate the risk of substances present in very low amount in food items (e.f. flavourings); nowadays it is used in many other sectors. SCCS, SCHER, SCENIHR, Joint Opinion on the Use of the Threshold of Toxicological Concern (TTC) Approach for Human Safety Assessment of Chemical Substances with focus on Cosmetics and Consumer Products, 8 June

28 Threshold of Toxicological Concern (TTC) TTC values have been derived by a statistical analysis of large toxicological data bases containing data on systemic toxicity reference values (NOEL, LOAEL, etc). Starting point: NOAELs distribution for different chemicals. The value corresponding to the 5 th percentile divided per 100 (uncertainty factor) gives rise to the TTC value 28

29 29

30 Cramer class I 30 µg/kg BW Cramer class II e III 1.5 µg/kg BW It is absolutely crucial to have available reliable data on the level of exposure. When the human exposure is below the TTC value, it is considered that the probability of experiencing any adverse effect is very low. 30

31 TTC Applicability domain 31

32 Pesticides relevant metabolites in groundwater Introduced for the first time in the previous Dir. 91/414/CEE. Metabolites are defined relevant when accounting for > 10% of the parent in environmental matrixes or are (eco)toxicologically relevant. The EU directive 98/83/CE (Quality of water for human consumption) indicates that the maximun tolerated levels for any pesticide or its metabolite should be µg/l (single and total). A Guidance Document on relevant metabolites in groundwater is available (EU Comm.,2003) establishing a procedure to be followed The evaluation is described as a step-wise procedure in 5 phases, following the general principle pf RA, but introducing the application of the TTC.

33 Phase 1 Starting from the analysis of all metabolites, some are excuded during phase 1: Metabolites with no or scant toxicological relevance (es.co 2 ), Inorganic compounds, Aliphatic organic compounds with <4 C chain lacking any eteroatoms and structural alert (es. epoxides, nitrosamine). These are defined as irrelevant metabolites (independently on their concetrations). The remaining ones proceed to phase II

34 Phase 2 Groundwater concentrations are considered (predictive models PEC gw or monitoring data) Metabolites with estimated or measured concentrations <0.1 g/l are excluded from further analysis as not relevant Phase 3 For those metabolites with estimated or measured concentrations >0.1 g/l toxicological data should be evaluated along three steps : not meeting only one out of the three steps requirement is sufficient to define the metabolite as relevant.

35 Step 1: comparison with the parent acute toxicity Toxicity parent relevant metabolite (<0.1 g/l) Toxicity <50% parentale step 2 Step 2: genotoxicity assessment. Genotoxic relevant metabolite (<0.1 g/l) Non genotoxic step 3 Step 3: toxicity assessment. T+, T, reprotox relevant metabolite (<0.1 g/l) carcinogen 1a,1b relevant metabolite (<0.1 g/l) carcinogen 2 relevant metabolite (<0.1 g/l), unless data can demonstrate the lack of relevance For the last case and remaining ones phase 4

36 When conc. is in the range µg/l a more detailed RA is required It is assumed as a general quality criteria for GW that the 10 µg/l should not be exceeded (this limit as no toxicological basis) and the expert should evaluated the situation on a case-by-case basis ANSWER Workshop 2017 Phase 4 The exposure level is compared with the TTC = 1.5 g/person/day (0.02 g/kg bw/d) Considering a 2L of drinking water daily intake, the maximum level acceptable in water is = 0.75 µg/l. When conc. < 0.75 µg/l no further evaluation When conc. > 0.75 µg/l phase 5 Phase 5

37 In vitro studies in RA Limited use for risk assessment purposes difficulties in carrying out quantitative in vitro to in vivo extrapolation (QIVIVE) translate in vitro effect concentration into human toxicologically equivalent dose Need of translating information from the cell level, to organs and subsequently to organisms and to distinguish between adaption vs. adversity, likely identifying actual in vitro markers of adversity (Blaauboer et al, 2012) or Key Events of AOPs Integrated approach: in silico and in vitro IATA Lack of information on actual cell exposure Battery KE (TD) + kinetics PBTK models in vitro biokinetics 37

38 Kinetics is finally considered the crucial body of information for the design and performance of traditional in vivo toxicological tests, toxicity data interpretation, identification of internal dose. Why not to include kinetics in alternative/non animal testing strategy? Biokinetics processes have been evoked to explain the in vitro/in vivo differences, but in vitro the nominal applied concentration rather than the actual level of cell exposure is usually associated to the observed effects. Figure from Heringa et al., ES&T,

39 In vitro biokinetics Test Item Evaporation Chemical instability Protein binding Passive/Active (Transporters) Plastic adsorption Uptake Cells Free Concentration in the medium Characterization of the cell model Target Metabolism Free intracellular Concentration 39

40 Adsorption to plastics and attachment to matrices Dependent on: Lipophilicity : LogD 7.4 >2.5 (e.g. amiodarone, CsA, Chlorpromazine) up to 70% plastic bound. Negligible binding for Ibuprofen, cisplatin, adefovir Time : increase with time of treatment Dose: increase with dose up to a plateau Serum competes with plastics Possibility of sequestration by Collagen; lower by Gelltrex 40

41 Binding to protein in the medium Cytotoxicity depends on protein binding in the medium Low albumin concentration High albumin concentration Cell uptake is reduced by protein binding in the medium Gülden M. et al. Toxicol. Letters 2003, 137,

42 It depends on: the cell type Dose time Accumulation in cells Transporter activity and Metabolic competence CsA Bellwon et al, TIV, 2015a Bellwon et al, TIV, 2015b Wilmes A., et al. Journal of Proteomics,

43 AOP : Pathways/sequence of biological perturbations/events leading to adverse effects MOA : a specific AOP is relevant for the chemical under evaluation? Verify the presence of key events Key Events in AOP: endpoints (readouts) to be measured to verify if the chemical acts according to a specific AOP. 43

44 OECD TG 442C Adopted: 4 February 2015 In Chemico Skin Sensitisation: Direct Peptide Reactivity Assay (DPRA) OECDTG 442D Adopted: February 2015 In Vitro Skin Sensitisation: ARE-Nrf2 Luciferase Test Method OECDTG 442E: Adopted: July 2016 In Vitro Skin Sensitisation: Human Cell Line Activation Test (h- CLAT)

45 OECD TG 442C : DPRA + sensitiser - OECDTG 442D : KeratinoSense + Sens. - OECD TG 442E : h-clat DPRA or protein reactivity: represents the initiating molecular event in the skin sensitization AOP. KeratinoSense TM or luciferases test provide data on the second key event in the AOP with induction of gene which are regulated by ARE (Antioxidant Response Element). h-clat readout represents the 3 rd key event in the AOP and quantify the expression changes of cellular surface markers associated to the monocytes and dendritic cells activation.

46 The issue of exposure to multiple chemicals Modified from A. Boobis,

47 The RA process up to now has been essentially carried out referring to single chemicals. In many cases it addressed a single route of exposure or refers to a single use (the same chemicals can be used in different sectors and the routes and exposure scenarios can be very complex. Exposomes approach and biomonitoring consider this opportunity (the internal dose sums up all the different contributions) What about exposure to multiple chemicals? What about multiple stressors exposure? Testing of mixtures is practically unfeasible (high number of components and combinations; variation in the environment due to bacterial metabolism; bio/photo and chemical degradation different for different chemicals, altering the relative content; variation over time)

48 48

49 Dose Additivity : the component-based approach When data on MoA are available: RPF (Relative Potency Factor) used for OPT and carbamates ; AChE inhibition as critical effect toward a reference compound (RPF=1) within the family: summing up the relative potency of single components the inhibitory activity of the mixture is obtained TEF (Toxic Equivalent Factor), similar to RPF, used for complex mixtures of dioxin like compounds (2,3,7,8,-TCDD is the reference compound for which TEF=1 and the critical effect is the binding to AhR receptor). The toxicity of the mixture is then obtained by summing up the product of specific RFP or TEF with the concentration of the single components. 49

50 Hazard Index When mechanistic data are not available : HI (Hazard Index) : assumption is that there is addivity as a worst case. The RfD or HBV of single components is needed. The relative contribution of each single is derived as the ratio between its concentration and the crresponding RfD. HI of the mixture is obtained by summing up single contributions : HI= Conc 1 /RfD 1 + Conc 2 /RfD Conc n /RfD n When HI>1 it is necessary to refine the RA on the basis of the expert judgement, since interactions (other than additivity) cannot be excluded 50

51 HazDaT Data base on line (ATSDR, 1997) containing data on the environmental contamination of >2000 sites on which ATSDR carried out an assessment aimed to protect public health Identification of more frequent combinations of environmental contaminants in water, soil, air, or in highly risky areas and description of interaction profiles (IP) IP for POPs in fish and in breast milk (dioxine-like chemicals, DDE, HCB, PCBs, Methyl-Hg); IP for 1,1,1-TCEthane; 1,1-DCEthane; TCE e PCE; IP for benzene, toluene, ettylbenzene e xilene (BTEX); IP for Cu, Pb, Mn e Zn and for Cd, As, Cr, and Pb IP for atrazine, simazine, desetilatrazine, diazinon and nitrates. 51

52 4 2 (L-Arginine) MCs acute hepatotoxic potential is congener-dependent Toxin i.p. LD 50 (µg/kg) M.W. Structure MC-LR cyclo -(D-Ala-L-Leu-D-MeAsp-L-Arg-Adda-D-Glu-Mdha-) [D-Asp 3 ]MC-LR cyclo -(D-Ala-L-Leu-D-Asp-L-Arg-Adda-D-Glu-Mdha-) MC-LA cyclo -(D-Ala-L-Leu-D-MeAsp-L-Ala-Adda-D-Glu-Mdha-) MC-YA cyclo- (D-Ala-L-Tyr-D-MeAsp-L-Ala-Adda-D-Glu-Mdha-) MC-YR cyclo -(D-Ala-L-Tyr-D-MeAsp-L-Arg-Adda-D-Glu-Mdha-) [Dha 7 ]MC-RR cyclo -(D-Ala-L-Arg-D-MeAsp-L-Arg-Adda-D-Glu-Dha-) MC-RR cyclo -(D-Ala-L-Arg-D-MeAsp-L-Arg-Adda-D-Glu-Mdha-)

53 In analogy with the method used for polychlorinated dibenzo[p]dioxins (PCDD), it has been proposed that one derive a toxicity equivalent factor (TEF) from the available acute toxicological data for MCs and NODs (Wolf and Frank, 2002; Funari and Testai, 2008). The approach should at present be limited to acute toxicity, since repeated toxicity data on different congeners are not available. The reference cyanotoxin is MC-LR, with TEF = 1; the TEF of a specific toxin (X) is derived as the ratio between the LD50 values, according to the equation: TEF X = LD50 MC-LR /LD50 X The total acute toxicity of the mixture is estimated by the sum of all the individual toxicity equivalents obtained as the product between the specific TEF and the toxin concentration. 53

54 By using this approach a more realistic assessment is obtained for a hypothetic mixture when compared with the worst case approach, considering all the component as toxic as MC-LR, which is generally higher. 54

55 International activity on mixtures: some examples Council Conclusion on Chemical Mixtures (2009) Kortenkamp et al. (2009) State of the Art on Mixture Toxicity. Report to the EU ECETOC (2011) Report Development of guidance for assessing the impact of mixtures of chemicals in the aquatic environment Meek et al. (2011) Risk assessment of combined exposure to multiple chemicals: A WHO/IPCS framework, Reg Tox Pharm SCHER, SCENIHR, SCCS (2012): Toxicity and Assessment of Chemical Mixtures Euromix Project funded by EU Workshop EFSA RIVM on mixture toxicty (Utrecht, 2016) CURRENT EFSA WG on MIXTURES to adopt a harmonised opinion on the issue 55

56 The EFSA Activity Panel on Plant Protection Products and their Residues EFSA (2008) suitability of existing methodologies assessing cumulative and synergistic risks from pesticides to human health to set MRLs (Regulation (EC) 396/2005). EFSA (2009) Risk assessment Cumulative Effects- Triazole fungicides EFSA (2012) Science behind the development of a risk assessment of Plant Protection Products on bees EFSA (2013) 1.Identification of pesticides to be included in cumulative assessment groups on the basis of their toxicological profile. 2.Relevance of dissimilar mode of action and its appropriate application for cumulative risk assessment of pesticides residues in food. Panel on Contaminants in the Food Chain EFSA (2008) Polycyclic Aromatic Hydrocarbons in Food EFSA (2009) TEF approach-non-ortho polybrominated biphenyls Marine biotoxins Saxitoxin Group Pectenotoxin Group EFSA (2011) Whole mixture approach applied to Mineral Oil Saturated Hydrocarbons EFSA (2012) dose addition approach-pyrrolizidine and Ergot alkaloids 56

57 The WHO framework 57

58 Start with exposure Use pattern Is exposure even possible? Physicochemical properties Is systemic exposure possible? Threshold of toxicological concern Is exposure so low that it can be ignored? If additivity is a potential concern, consider comparing exposure with a fraction of the TTC Health based guidance values If exposure to a single chemical exceeds its HBGV (e.g. TDI), address this issue before Cumulative RA Modified from A. Boobis,

59 Modified from J-L Dorne,

60 In silico Kinetics omics Imaging organ on a chip Modified from A. Boobis,

61 61