Approaches to Additivity Thomas Backhaus, University of Gothenburg Email: thomas.backhaus@gu.se Twitter: @ThoBaSwe

Approaches to mixture assessment Whole mixture testing Sophisticated PBPK models Lead-compound approach Component-based approaches

Lead compound approach One defined compound is selected as the “lead” compound of a given (sub)mixture It is assumed that the whole (sub)mixture has a toxicity similar to this compound That is, the mixture is simply assumed to comprise only the lead compound Used in CLP, REACH

Approaches to mixture assessment Whole mixture testing Sophisticated PBPK models Lead-compound approach Component-based approaches

Approaches to Additivity Effect Summation Response Addition Concentration Addition

Adding Effects Effect Summation E(mix) Effect of the mixture E(ci) Effect of compound if applied singly ci Concentraion of compound i n number of mixture components

Effect Summation leads to inconsistent assessments

Effect Summation leads to inconsistent assessments

Approaches to Additivity – Adding probabilities Response Addition, Independent Action Mortality Survival E(mix) Effect of the mixture E(ci) Effect of compound if applied singly ci Concentraion of compound i n number of mixture components

Approaches to Additivity – Adding probabilities Response Addition, Independent Action Joint effect is higher than each individual effect! Subst.1 = 50% Subst.2 = 50% Mixture = 75%

Adding fractions of equi-effective concentrations Concentration Additivity, Toxic Unit Summation, Additivity formula RQ Risk quotient of the mixture ECx(mix) Concentration of the mixture that causes x% effect cMix Concentration of the mixture ECxi Concentration of compound I that causes x% effect if applied singly ci Concentration of compound i in a mixture that gives x% effect n number of mixture components

Approaches to Concentration Additivity Simple similar action Toxic Equivalency Factor Relative Potency Factor Point of Departure Index Hazard Index Addition of risk quotients (PEC/PNEC’s)

Simple similar action

Simple similar action Simple Similar Action, Relative Potency Factor, Toxicity Equivalency Factor, Toxic Equivalent Quantity (TEQ) TEFi Toxicity Equivalency Factor for compound I ci Concentration of compound i in the mixture n number of mixture components

Not-so-simple similar action Faust et al. 2001. Predicting the joint algal toxicity of multi-component s-triazine mixtures at low-effect concentrations of individual toxicants. Aquatic Toxicology

Not-so-simple similar action Altenburger et al. Predictability of the toxicity of multiple chemical mixtures to Vibrio fischeri: mixtures composed of similarly acting chemicals. Environmental Toxicology and Chemistry

Not-so-simple similar action Altenburger et al. Predictability of the toxicity of multiple chemical mixtures to Vibrio fischeri: mixtures composed of similarly acting chemicals. Environmental Toxicology and Chemistry

Point of Departure Index Point of Departure Index (PODI) Endpointi NOEC of compound i recorded in a specific assay PODi Point of Departure for compound i

Point of Departure Index Point of Departure Index (PODI) Endpointi NOEC of compound i recorded in a specific assay PODi Point of Departure for compound i ci Concentraion of compound i n number of mixture components

Hazard Index Hazard Index, PEC/PNEC summation PNECi Predicted No Effect Concentration of compound i PNECMixture Predicted No Effect Concentration of the mixture i PECi Predicted Environmental Concentration of Compound i PECmixture Predicted Environmental Concentration of the mixture n number of mixture components

Hazard Index – Difficult to interpret Compound 1: PEC1=0.4*10-4 EC50Algae: 1.0 EC50Daphnids: 0.1 PNEC = 10-4 EC50Fish: 1.0 Compound 2: PEC2= 0.8*10-4 EC50Algae: 0.1 PNEC = 10-4 EC50Daphnids: 1.0

Hazard Index – Difficult to interpret Compound 1 Compound 2 0.4 0.8 ?

The following slide was added as a consequence of the discussions during the workshop. For further details on the suggested tiered approach for using additivity in the context of REACH (and similar frameworks), please see: Backhaus, T., Faust, M. Predictive environmental risk assessment of chemical mixtures: a conceptual framework, Environmental Science and Technology, 46(5), 2012, http://dx.doi.org/10.1021/es2034125

CA needs to be applied in a tiered fashion CA using PNEC values CA using actual toxdata for each concerned species & endpoint Comparative application of CA and IA using full concentration-response functions toxdata for each concerned species & endpoint

Commonalities of the different CA incarnations Component-based Mixture risk is higher than risk of individual compounds Need to know exposure level (internal or external) Need to have a potency measure that relates to the same level of effect

Commonalities of the different CA incarnations Knowledge needed on the underlying modes of action Data demands Ease of interpretation

Critical issues (in the context of this workshop) Mixtures might be chemically poorly defined Synergistic and/or antagonistic interactions Handling of poorly soluble but still toxic compounds

CA in the context of UVCB substances CA is component-based and can thus only be applied to defined mixture Use in the context of UVCBs: application to the defined part in order to identify important mixture drivers, assess the (eco)toxicological impact of batch-to-batch variations, etc. Used in TIE (Toxicity Identification and Evaluation) resp. EDA (Effect Directed Analysis) to confirm identified components and mixture drivers.

CA in the context of UVCB substances Brack. Effect-directed analysis: a promising tool for the identification of organic toxicants in complex mixtures? Anal Bioanal Chem (2003)

Synergistic or antagonistic interactions Synergism: toxicity is higher than expected by toxic unit summation, i.e. less of a mixture is needed to cause a predefined toxicity. Antagonism: toxicity is lower than expected by toxic unit summation, i.e. more of a mixture is needed to cause a predefined toxicity. Can be cause by toxicodynamic, toxicokinetic interactions, and/or by data gaps and biases.

A real-world pesticide mixture Decreasing TU Finizio et al., Agr. Eco. Env., 111,111-118, 2005 Junghans et al., Aquatic Toxicology, 76, 93-110, 2006

A real-world pesticide mixture 1 compound 10x more toxic than estimated

A real-world pesticide mixture First n compound 10x more toxic than estimated

A real-world pesticide mixture n randomly selected compounds 10x more toxic than estimated

The implications of CA-data demands

The implications of CA-data demands

The implications of CA-data demands

The implications of CA-data demands

The implications of CA-data demands

Final comments and conclusions “Additivity” can mean whole lot of different things… “Concentration Additivity can mean a whole lot of different things Most common incarnation: toxic unit summation, hazard index Mixtures buffer against synergistic interactions – the more compounds, the better

Final comments and conclusions CA does not directly allow to calculate the expected effect of a mixture. Full concentration-response curves for each component are needed for that purpose. Applicability is limited to the effect range that is covered by all compounds. Problem if some compounds do not reach a pre- defined effect level.

Final comments and conclusions Simplifications of CA (Hazard Index etc) make data collection easier and interpretation more fuzzy.

Final comments and conclusions CA, mainly in the form of toxic unit summation or hazard indices, is the only mixture toxicity concept that has found widespread application (REACH, CLP, but also pesticide & biocide regulations). Detailed mode-of-action based approaches often discussed and included in the guidelines, but rarely used. Lack of data, lack of concern (motivation).

Approaches to Additivity Thomas Backhaus, University of Gothenburg Email: thomas.backhaus@gu.se Twitter: @ThoBaSwe