Threshold of Toxicological Concern an approach for safety assessment and its applicability to cosmetics-related chemicals. July 24, 2014 Chihae Yang

Transcription

1 Threshold of Toxicological Concern an approach for safety assessment and its applicability to cosmetics-related chemicals July 24, 2014 Chihae Yang

2 Overview Decisions Next steps I. Brief history of TTC III. TTC approach II. Chemicals in cosmetics products

3 Threshold of toxicological concern Concept pioneered by US FDA Center for Food Safety & Applied Nutrition (CFSAN) Frawley 1967 ( 220 chemicals, chronic toxicity) Rulis 1986 ( 500 chemicals, US FDAPAFA database) Cheeseman 1999 (709 chemicals RTECS and CPDB) Threshold dose below which no relevant effects are expected, regardless of the substance Applicable to low amount or low exposure substances for which limited toxicity data are available The dose makes the poison!

4 Threshold of regulation (TOR) exemptions Food contact material (Part in 21 CFR) Estimated consumer exposure not to exceed 0.5 ppb in the daily diet No evidence of carcinogenicity in humans or animals No structural basis for suspecting it to be a carcinogen or a potent toxin The first TOR in 1996: tetra sodium EDTA (information provided by Kirk Arvidson, US FDA CFSAN, OFAS) /ThresholdRegulationExemptions/default.htm require compound specific data 0.5 ppb (1.5 mg) not expected to have appreciable risk to human health 4

5 Cancer TTC Data and endpoints considered in FDA analyses: Chronic toxicity (US FDA PAFA and RTECS databases) Cancer potency (CPDB database) Cancer potency POD: TD substances (1999) Carcinogenicity is the endpoint of most concern at the lowest dietary concentrations. Sue Barlow, 2005 Threshold at mean virtual safe dose (one in million cancer risk) Structural alerts: N-nitroso, organophosphorus, strained heteronuclear rings, heavy metals, alphanitro furyl, hydrazine/triazines/azides/ azoxy, polycyclic amines, Cheeseman et. al., FCT, 37, 387, Kroes et. al, FCT, 42, p62, Barlow S, et. al. 2005, ILSI Europe Concise Monograph Series 5

6 Non-cancer TTC Data and endpoints considered by Munro et. al: Chronic toxicity US EPA IRIS, JECFA, literature) Non-cancer endpoints: POD: NOEL 613 substances (1996) Generic TTC method by correlation of structural classes with NOEL III I Structural classes: Cramer I, II, III (1978 publication) Sue Barlow, 2005 Fifth percentile threshold for each Cramer Class Munro et. al. Food Chem Tox 34, p ,

7 Cramer Decision Tree Chemical classification scheme (for defined organics) to predict toxicity Class I: suggests low oral toxicity ( innocuous ), simple structures with efficient metabolism Class II: less innocuous but no structural features suggesting toxicity Class III: not strongly presumed to be safe; may suggest toxicity or have reactive functional groups is aromatic? c v have N-nitroso, azo, triazeno, quat.? Have one sulfonate or sulfonamide per 20 or fewer carbons? Food Cosmetics Toxicol 16, p , 1978 c v c v 7

8 Correlation of structure classes with No Observed Effect Levels Cramer 1978 Munro 1996 NEL (mg/kg/day) N NOEL(mg/kg/day) N Class I Class II Class III Cramer GM et. al. Food Cosm. Tox 16, p255, Estimation of toxic hazard a decision tree approach

9 Cancer and non-cancer combined Cancer threshold proposed: 1.5 microgram/day. Non-cancer thresholds proposed by Munro analysis Cramer 5 th % NOEL mg/kg/day Human exposure mg/person/day I II III Kroes R et al. Food Chem. Toxicology 42 (2004) = / with safety factor of 100

10 A simple example 2,4 and 2,5-xylenols have been reported with skin tumors (dermal studies). 2.6-xylenol is reported negative in NTP salmonella assays. Chemical information Name 2,6-xylenol CAS RN Cramer Class CEDI* COSING chem func. Class I (by Munro and Toxtree v2.6) 9.7 ppb mg/kg/day Perfuming agent *FDA CFSAN CEDI (Cumulative Estimated Daily Intake) Database:

11 A simple example 2,6-xylenol for a cause where the use level exposure at 1 mg/day assumed non-mutagenic (Ames negative by NTP) OP? Cramer Class? No EDI<1800 mg/day? Class I Yes expected to have a negligible risk from this case Kroes, et. al

12 Overview Decisions Next steps III. COSMOS TTC II. Chemicals in cosmetics products I. Brief history of TTC

13 Requirements of non-cancer TTC approach Structural Classes Chemical space Food Cosmetics Toxicol 16, p , 1978 NOEL/NOAEL database Tox data NOEL/NOAEL LOEL/LOAEL Pragmatic TTC approach Munro, et. al. Food Chem. Tox 34 (1996)

14 Chemical space Chemical space TTC Concept developed from oral exposure studies primarily for food contact materials assessment Does Munro dataset cover typical cosmetic ingredients and impurities? Blackburn et. al a feasibility study based on a small personal care products (Reg. Tox, Pharm. 43,249, 2005) A. Worth et. al Analysis of a preliminary COSMOS TTC dataset (EUR EN, 2012)

15 Cosmetics Inventory A reference (look-up) list for defining substances used in cosmetics products Cosmetics-related chemicals Voluntary Cosmetics Registration Program Inventory 19,587 INCI names 9,857CAS RNs FDA VCRP list is provided by US CIR (2013, 2014) Compilation of Ingredients Used in Cosmetics in US, JE Bailey, Ed. 2010

16 COSMOS DB v1.0 Cosmetics Inventory in v1.0 INCI names (17,101) from EU COSING and US VCRP Curated 5,500 structures Over 2,200 structures in common with FDA PAFA (legacy database) COSING chemical function VCRP product category Database searching *Initial structures: FDA, DSSTox, P&G Tools & workflow Data entry Total of 81,602 compound records

17 COSMOS Inventory in COSMOS DB v1.0 connect to COSING and SCCS opinion INCI name 17

18 COSMOS Inventory in COSMOS DB v1.0 connect to COSING and SCCS opinion INCI name Cosing chem functions 18

19 Cosmetics Inventory COSING chem function VCRP (voluntary cosmetics registration program) product category SKIN CONDITIONING,PROTECTING EMULSIFYING, EMULSION STABIL,SURFACTANT PERFUMING HAIR DYEING, COSMETIC COLORANT HAIR CONDITIONING EMOLLIENT ANTIOXIDANT AM, PRESERVATIVE UV FILTER, ABSORBER PLASTICISER % of all chem functions (log scale)

20 Cosmetics Inventory Chemical Space ToxPrint chemotypes Cosmetics inventory v1.0 alcohol alcohol, phenol alochol, diols aldehyde amine amine, aromatic (NH2) azo halide ketone ketone, ACAC carboxylic ester organometal phosphorus pyran, generic silicon steroid sulfide sulfonyl group urea aliphatic chain >= C8 non-ionic surfactant anionic surfactant cationic surfactant - QUAT Hair dye Perfume, Fragrance Emulsifying agent, Emulsion stabilizer, Surfactant % in collections (in Log scale)

21 Compare Cosmetics and Food ToxPrint chemotypes Cosmetics/PAFA Total Food and cosmetics broadly share similar chemical space. - Comparisons were made after removing cosmetics from PAFA structures. Quat cationic surfactant Non-ionic surfactant

22 Overview Decisions Next steps II. Chemicals in cosmetics products I. Brief history of TTC III. COSMOS TTC

23 Requirements of non-cancer TTC approach Structural Classes Cosmetics Inventory NOEL/NOAEL database Tox data NOEL/NOAEL LOEL/LOAEL Pragmatic TTC approach

24 Issues in data availability Regulatory requirements Few cosmetics ingredients are subject to EU regulation e.g., coloring agents, preservatives and UV filters Different levels of toxicity testing application, substance types repeated dose toxicity data are often not available for minor components or impurities Proprietary data Immediate availability for safety evaluation NOEL/NOAEL database

25 Issues in oral vs. dermal data 1. Develop dermal TTC based on reliable dermal repeated dose toxicity data 2. Extend the oral TTC by applying an oral-to-dermal extrapolation approach Cosmetics Inventory Oral TTC (current approach) Primary exposure route of cosmetics products is dermal Oral toxicity data NOEL/NOAEL LOEL/LOAEL Kroes et. al 2007 Feasibility study of oral-to-dermal extrapolation including bioavailability based on Munro dataset Food Chem. Tox 45 (2007)

26 COSMOS Project and TTC Expert Groups ILSI Europe Expert Group 1: Extend oral TTC dataset with cosmetics-related chemicals Review reliability and relevance of selected NOAELs Toxicity assessment (ILSI EWG1/FDA) Bioavailability assessment (ILSI EWG2/COSMOS) Combined strategy ILSI Europe Expert Group 2: Estimate exposure differences between dermal and oral delivery Address metabolism differences between skin and liver 26

27 Curation strategy of COSMOS TTC dataset 1. Toxicity database 2. NOAEL database 3. TTC dataset Study inclusion criteria Study inclusion criteria Study relevance NOAEL selection criteria NOAEL decision Filter Munro 1996 Filter orepeattox DB NOAEL database TTC dataset V1.7 current interim version All compounds in COSMOS TTC dataset have toxicity data either in orepeattox DB v1.0 or data in FDA PAFA, or EPA ToxRefDB unless the compounds are in either Munro or EPA IRIS. 7 iterations of dataset evaluation and 2 QC sessions of study reviews

28 COSMOS orepeattox DB - Oral repeated-dose toxicity database * orepeattox DB Public *ECHA: REACH Substance Registration Database SCCS: Published opinions from EC Scientific Committee of Consumer Safety Manually harvested ( ) Original publication/documents were harvested whenever available. ToxRefDB entry tool adopted for COSMOS 228 chemicals 186 Cosmetics-related chemicals 100 hair dyes 42 impurities (CERES originated) 340 oral toxicity studies chronic, repro-dev, subacute with 28 days, subchronic studies Black dotted line: manual harvesting 28

29 COSMOS addition orepeattox DB

30 COSMOS orepeattox DB and minimum study (MINIS) inclusion criteria Must meet: Subacute, subchronic, chronic, reproductivedevelopmental, carcinogenicity (no neoplastic lesions) Must meet rat, mouse, dog, monkey rabbit (only for DART studies) Must meet COSMOS MINIS criteria Dose groups, # of animals, days/week treatment Species Dose and regimen Study type Oral exposure Effects findings at dosegroup Duration Reference Data sources Must meet 28 days (except DART studies) Must meet All original documents when ever available (except regulatory study reports) Must meet COSMOS MINIS criteria All effects are recorded (neoplastic lesions are not included)

31 Power of ontology-driven database: orepeattox DB profiles Most sensitive species rat O O O Most common organs: liver, kidney, stomach, forestomach, spleen, lung, thyroid, heart Most common lesions for liver inflammation, fatty change, cellular infiltration, cytoplasmic vacuolation O N O N O Cl O O O N N O O Cl O O Cl O O O Cl N O O O Liver phenotypes/tissue, cells, organelles 31

32 Data filters in the curation of COSMOS TTC dataset 1. Toxicity database 2. NOAEL database 50% of the data has been manually harvested by COSMOS, FDA CERES, or EPA ToxRefDB 3. TTC dataset Study inclusion criteria of orepeattox DB are applied to all studies from other data sources (FDA PAFA or ToxRefDB). NOAEL selection criteria Lowest NOAEL with clear LOAEL Free Standing NOAELs were excluded whenever possible Chronic studies were preferred when availble Guideline studies were preferred NOAEL QC (2 QCs by expert review sessions) Lowest 10% of each Cramer Class or the compounds whose NOAEL values greatly differ across various data sources Toxicologically meaningful and human relevant when mechanism is known.

33 Description of COSMOS TTC v1.7 (on-going) Compound descriptions COSMOS TTC v1.7 (558) Munro* (607) Cosmetics Inventory Cramer Class I: Class II: Class III 239: 36: : 28: 443 Nutrients - Lipid soluble vitamins - Essential amino acids Compound classes - Hair dyes - Parabens - Phthalates A, D, E, K removed removed retinol phenylalanine

34 COSMOS TTC v1.7 vs. Munro TTC dataset Chemotypes COSMOS:Munro: Munro-cos Total alcohol 274 amines 115 alcohol_phenol 114 alcohol_diol 83 amine(nh2)_aromatic 19 ketone 61 sulfonyl 39 halide_organo 36 carboxylicester_aromatic 78 amine_ethanol* 13 pyran_generic 45 aliphatic chain >= C8 67 azo 13 aldehyde 28 quatn 20 nitro_aromatic* 9 phosphorous_organo 12 urea 10 sulfide 3 silicon 1 steroid ring 3 COSMOS has a more diverse physchem properties profile. *Cosmetics in Munro are not likely hair dyes, but many are anionic surfactants.

35 Issues being addressed Cramer Class assignment conflicts between Munro and Toxtree (v2.6) have been resolved manually. Unreliable data within 5 th percentile (Class I): data for isopropyl alcohol, ascorbic acids, 4-butyrolactone have been replaced by more reliable studies during additional QC. Parabens and phthalates: Munro in general less conservative than COSMOS NOAEL database. These substances affect the 5 th percentile of the Cramer Class I (propyl and butyl paraben and dibutyl phthalate). EG1 is currently evaluating. Hair dyes: Most of the 110 hair dyes are Class III compounds. However they do not affect the 5 th percentile when removed. Cramer Class II: not a viable group. Combine with Class III.

36 Next steps planned Finalize the COSMOS TTC dataset by the end of Q and port to COSMOS DB Finalize the skin permeability database by the end of Q and port to COSMOS DB Develop TTC tools (planned for delivery in 2015 June) Enter query compound Assign Cramer Classe and compound categories Check the chemical applicability domain Apply TTC decision tree including the bioavailability tree developed from Expert Group 2

37 Acknowledgements ILSI Europe Expert Group 1 A Boobis, S Felter, H Hollnagel, K Jacobs, A Worth, C Yang QC1: R Safford QC2: K Blackburn, E Rufer ILSI Europe Expert Group 2 G Barrett, M Cronin, R Guy, J Plautz, A Perry, N Monteiro- Riviere, C Roper, H Rothe, D Rua, F Williams, J Westerhout, C Yang Sue Barlow ILSI Europe COSMOS partners V Vitcheva, D Keller, K Arvidson M Checheva, E Fioravanzo, M Gajewska, D Hristozov, J Madden, A Mostrag-Szlichtyng, M Nelms, JF Rathman, A. Richarz, C Schwab, F Steinmetz, L Terfloth, I tsakovska US EPA NCCT Ann Richard, Matt Martin US CIR (US VCRP list) I Boyer, B Heldreth US FDA CFSAN CERES team Kirk Arvidson (CERES DB) P&G (COSMOS DB initial structures) Joan Fischer, Mike Laufsweiler, others

38 Acknowledgements European Community s Seventh Framework Program (FP7/ ) COSMOS Project under grant agreement n and from Cosmetics Europe. D A T A A T A D