Mike Comber Consulting TIMES-SS Assessment of skin sensitisation hazard Presented on behalf of the TIMES-SS consortia
Outline Why TIMES-SS? Why TIMES-SS? TIMES-SS v1 TIMES-SS v1 - Brief overview of the model - Brief overview of the model Performance & issues Performance & issues The need for further work - TIMES-SS v2 The need for further work - TIMES-SS v2 Consortium & Aims Consortium & Aims Current issues Current issues
Why TIMES-SS? There are in-vitro alternatives & other models….. There are in-vitro alternatives & other models….. But no one model is helpful on its own But no one model is helpful on its own TIMES-SS attempts to be mechanistic TIMES-SS attempts to be mechanistic Improves understanding Improves understanding More acceptable to regulators and toxicologists More acceptable to regulators and toxicologists LMC are very open to development LMC are very open to development TIMES-SS is available to the consortia during the development of v2 TIMES-SS is available to the consortia during the development of v2
Aims for TIMES-SS To develop a skin sensitisation (Q)SAR model that: To develop a skin sensitisation (Q)SAR model that: Potentially minimises the need for animal testing Potentially minimises the need for animal testing Is scientifically credible and valid to Industry and Regulatory bodies Is scientifically credible and valid to Industry and Regulatory bodies Agrees with the OECD principles for (Q)SAR validation Agrees with the OECD principles for (Q)SAR validation Mechanistically defensible Mechanistically defensible Hence has high potential for acceptance under REACH in place of animal tests Hence has high potential for acceptance under REACH in place of animal tests
TIMES-SS v1 TIssue MEtabolism Simulator TIssue MEtabolism Simulator Estimates skin sensitisation potency using a simulator for skin metabolism with (Q)SARs. Estimates skin sensitisation potency using a simulator for skin metabolism with (Q)SARs. Training set of 729 chemicals with experimental data from three sources (LLNA, GPMT, BgVV). Training set of 729 chemicals with experimental data from three sources (LLNA, GPMT, BgVV). Predicts potency as one of three classes: significant, weak or non sensitising. Predicts potency as one of three classes: significant, weak or non sensitising.
TIMES-SS v1 Skin metabolic simulator contains over 236 hierarchically ordered spontaneous and enzyme controlled reactions. Skin metabolic simulator contains over 236 hierarchically ordered spontaneous and enzyme controlled reactions. Covalent interactions of chemicals/metabolites with skin proteins are described by 47 alerting groups. Covalent interactions of chemicals/metabolites with skin proteins are described by 47 alerting groups. A multi-step applicability domain is incorporated into the model. A multi-step applicability domain is incorporated into the model.
External validation exercise Applicability domain of TIMES-SS defined Applicability domain of TIMES-SS defined Identified EINECS chemicals that fell within TIMES domain (6000 out of EINECS (60311)) Identified EINECS chemicals that fell within TIMES domain (6000 out of EINECS (60311)) “Randomly” chose 40 chemicals to test in the LLNA (12 predicted sensitisers & 28 predicted non-sensitisers) “Randomly” chose 40 chemicals to test in the LLNA (12 predicted sensitisers & 28 predicted non-sensitisers) Tested blind Tested blind Observed skin sensitisation effect was compared with TIMES model prediction. Results were evaluated in light of reaction chemistry principles. Observed skin sensitisation effect was compared with TIMES model prediction. Results were evaluated in light of reaction chemistry principles.
Evaluation For each of the test results For each of the test results TIMES-SS predicted a result, but also TIMES-SS predicted a result, but also Dave Roberts - expert assessment also used Dave Roberts - expert assessment also used Evaluated each of the data points where Evaluated each of the data points where TIMES-SS differed from the data TIMES-SS differed from the data Using the DR expert assessment Using the DR expert assessment Assessed on mechanistic grounds Assessed on mechanistic grounds
Characterisation with the OECD principles OECD principles for (Q)SAR validation: OECD principles for (Q)SAR validation: a defined endpoint a defined endpoint an unambiguous algorithm an unambiguous algorithm a defined domain of applicability a defined domain of applicability appropriate measures of goodness-of-fit, robustness and predictivity appropriate measures of goodness-of-fit, robustness and predictivity a mechanistic interpretation a mechanistic interpretation where demonstrates the concordance between TIMES and the OECD principles For full evaluations - see : Patlewicz et al., 2007 Reg Tox Pharm, 48, 225–239 & Roberts et al., 2007, Chem Res Toxicol, 20 (9), 1321–1330
Conclusions – TIMES-SS v1 Development of a QSAR with external validation to meet OECD principles Development of a QSAR with external validation to meet OECD principles 40 chemicals tested resulting in new LLNA data 40 chemicals tested resulting in new LLNA data The results were promising (initial concordance 75%) The results were promising (initial concordance 75%) Extensive evaluation to assess results in light of reaction chemistry principles Extensive evaluation to assess results in light of reaction chemistry principles To check hypotheses, 4 further compounds were tested To check hypotheses, 4 further compounds were tested The insights derived from all 44 compounds have helped to define: The insights derived from all 44 compounds have helped to define: short-term modifications/refinements for TIMES-SS short-term modifications/refinements for TIMES-SS mid-long term targets for new research work mid-long term targets for new research work
TIMES-SS v2 :Consortium Research team Research team Laboratory of Mathematical Chemistry, University Bourgas Laboratory of Mathematical Chemistry, University Bourgas Dr Dave Roberts Dr Dave Roberts Consortium Consortium ExxonMobil ExxonMobil Procter & Gamble Procter & Gamble Research Institute for Fragrance Materials Research Institute for Fragrance Materials Unilever Unilever Danish National Food Institute Danish National Food Institute ECB - JRC ECB - JRC Funding & data sharing + sweat equity Funding & data sharing + sweat equity
TIMES-SS v2 Under the auspices of the International QSAR Foundation Under the auspices of the International QSAR Foundation 9 milestones agreed for a 3 year programme 9 milestones agreed for a 3 year programme Milestone 1 – Identify a process for dealing with the test/training set chemicals which are identified out of domain Milestone 1 – Identify a process for dealing with the test/training set chemicals which are identified out of domain Milestone 2 – Address chemicals where data conflicts Milestone 2 – Address chemicals where data conflicts Milestone 3 – Provide guidance on how to address charged molecules Milestone 3 – Provide guidance on how to address charged molecules Milestone 4 – Assess the categories of sensitisation (strong/weak/non) Milestone 4 – Assess the categories of sensitisation (strong/weak/non) Milestone 5 –Implement modifications for inclusion in TIMES-SS v2 Milestone 5 –Implement modifications for inclusion in TIMES-SS v2 Milestones 6/7 – Assess missing/inaccurate mechanisms via literature Milestones 6/7 – Assess missing/inaccurate mechanisms via literature Milestone 8 – Gather new LLNA data for TIMES-SS v2 model. Milestone 8 – Gather new LLNA data for TIMES-SS v2 model. Milestones 9/10 – Assessment of abiotic/protein reactivity Milestones 9/10 – Assessment of abiotic/protein reactivity
Issues being addressed Improving TIMES-SS Improving TIMES-SS How to deal with conflicting test data How to deal with conflicting test data Within test systems & across test systems Within test systems & across test systems Using expert judgement to build/refine the model Using expert judgement to build/refine the model Including new or refined mechanistic descriptions for reactions Including new or refined mechanistic descriptions for reactions How to describe chemical reactivity in a model How to describe chemical reactivity in a model Providing the literature support Providing the literature support Are they better ways of describing reactivity and incorporating the information in TIMES-SS? Are they better ways of describing reactivity and incorporating the information in TIMES-SS?
Identifying the cause of an effect Complex toxicological endpoints are biological responses to: Complex toxicological endpoints are biological responses to: Direct molecular interactions dependent on chemical structure Direct molecular interactions dependent on chemical structure Indirect molecular interactions which are dependent on the chemical structure of metabolites Indirect molecular interactions which are dependent on the chemical structure of metabolites AND Biological processes dependent on other properties e.g. pH/chemical reactivity AND Biological processes dependent on other properties e.g. pH/chemical reactivity In TIMES-SS - trying to separate the unknowns associated with metabolism from the unknowns associated with chemical reactivity itself. In TIMES-SS - trying to separate the unknowns associated with metabolism from the unknowns associated with chemical reactivity itself. There has been a continual effort to make sure any plausible mechanism of interactions leading to a protein adduct is consistent: There has been a continual effort to make sure any plausible mechanism of interactions leading to a protein adduct is consistent: With the literature on that specific reaction mechanism, and With the literature on that specific reaction mechanism, and With the more general understanding chemical reactivity. With the more general understanding chemical reactivity.