Predictive Toxicology for Greener Chemicals

Craig Rowlands, PhD, DABTSenior Scientist for InnovationUnderwriters Laboratories

UL | Empowering Trust in a Complex World

Global Chemical Market

>130x106

Chemical structures synthesized

140,000Industrial chemicals used in global commerce

10,500chemicals used in personal care products

4,500Food Additives chemicals currently in use

Important data gaps may exist regarding chemical use and human & environmental health impact!

Chemical Compliance Pressures are Increasing

Dramatic increases in safety testing due to European REACH Registrations

- 2010 & 2013 → testing proposals- May 2018 → full submissions- ≥ 30,000 chemicals

New or changing regulations- Korea-REACH, Taiwan-REACH, China-REACH- Lautenberg Chemical Safety Act (new TSCA)

Other assessment/testing needs- Endocrine disruptors- Mixtures- Nanomaterials….

3

Chemical Sustainability Compliance Pressures are Increasing

• chemicals alternatives assessment (CAA) to substitute for the use of hazardous chemicals in products.

• >100 tools for comparing hazard characteristics of different chemicals.

• selection of alternatives for:• banned or restricted chemicals or

materials;• chemicals that are perceived as

hazardous by the public;• developing environmentally preferred

products.

The new regulator

CHEMICAL SUPPLY CHAINS ARE COMPLEX

sustainable productscompliant products

product safety (SDS)

innovate products

better ingredients

5

Need for Hazard ata

Can conventional toxicology approaches keep pace with the demands for chemical hazard data?

picture© ChemSec

$$$$$$$$$$$$$$$$$$$$$$

R&D Regulatory toxicology(e.g. REACH, EPA …)

Follow-up MoA(e.g. human relevance)

Example - Two-generation reproductive toxicity study:LAnimal-intensive (>3,000 animals)LExpensive (>$500,000) LSlow (15 months)

Minimal involvement of Toxicologist inproduct designor selection

Conventional Toxicology

7

New Safety Assessment Paradigm

$$$$$$$$$$$$$

R&D Toxicity testing,risk assessment

Early safety prediction

Moleculardesign

Selection

Read-acrossOptimize

formulations

Exposure assessment

Informtesting

Defense of commercialized or legacy molecules

8

Predictive Toxicology Integrated Technology Platforms

9

In Vitro Biological Profiling

Cheminformatics Exposure modeling

Cell cultures Zebrafish Genomics

PREDICT SAFETY

In Vitro Assays for Bioprofiling

Ex: Assays for Skin Sensitization*

Wong et al Front Pharmacol 6: 1-13, 2015

Class Test methodsCorrosion In vitro Skin Corrosion: Transcutaneous Electrical Resistance Test Method (TER): TG430

In vitro Skin Corrosion: Reconstructed Human Epidermis (RHE) test method: TG431

CORROSITEX Skin Corrosivity Test: TG435

Skin irritation In vitro Reconstructed Human Epidermis (RhE) Test methods, EpiDerm, EPISKIN, SkinEthic,LabCyte EPI-Model: TG439

Phototoxicity 3T3 NRU Phototoxicity Test: TG432

Eye irritation Bovine Corneal Opacity and Permeability Test Method for Identifying i) Chemicals InducingSerious Eye Damage and ii) Chemicals Not Requiring Classification for Eye Irritation or Serious Eye Damage: TG437Isolated Chicken Eye Test Method for Identifying i) Chemicals Inducing Serious Eye Damage and ii) Chemicals Not Requiring Classification for Eye Irritation or Serious Eye Damage: TG438

Fluorescein Leakage Test Method for Identifying Ocular Corrosives and Severe Irritants: TG460Short Time Exposure In Vitro Test Method for Identifying i) Chemicals Inducing Serious EyeDamage and ii) Chemicals Not Requiring Classification for Eye Irritation or Serious EyeDamage: TG491Reconstructed human Cornea-like Epithelium (RhCE) test method for identifying chemicalsnot requiring classification and labelling for eye irritation or serious eye damage: TG492

Skin sensitization In Chemico Skin Sensitization, Direct Peptide Reactivity Assay (DPRA): TG442CIn Vitro Skin Sensitization, ARE-Nrf2 Luciferase Test Method: TG442D

Endocrine activity screening

Performance-Based Test Guideline for Stably Transfected Transactivation In Vitro Assays toDetect Estrogen Receptor Agonists and Antagonists: TG455

H295R Steroidogenesis Assay :TG456BG1Luc Estrogen Receptor Transactivation Test Method for Identifying Estrogen ReceptorAgonists and Antagonists: TG457Performance-Based Test Guideline for Human Recombinant Estrogen Receptor (hrER) In VitroAssays to Detect Chemicals with ER Binding Affinity: TG493

Genotoxicity Bacterial Reverse Mutation Test : TG471In vitro Mammalian Chromosome Aberration Test: TG473In Vitro Mammalian Cell Gene Mutation Tests using the Hprt and xprt genes: TG476In vitro Micronucleus Test: TG487In Vitro Mammalian Cell Gene Mutation Tests Using the Thymidine Kinase Gene: TG490

Skin absorption Skin Absorption: In vitro Method: TG428

OECD Test Guidelines for in vitro Test Methods

Cheminformatics QSAR methodologies have been developed

• Cheminformatics is the integration of all relevant information on a compound and its structural or substructural analogs in order to make a preliminary assessment of potential toxicity

• Statistical – Predictions are made using a multilinear regression models fit to the training set using an algorithm-based approach.

• Mechanistic- Predictions are made using a multilinear regression model for toxicity modes of action associated with specific chemical features.

• Nearest neighbor method – The predicted toxicity is estimated by taking an average of a predefined number of chemicals in the training set that are most similar to the test chemical.

• Consensus method – The predicted toxicity is estimated by taking an average of the predicted toxicities from each of the above QSAR methodologies.

Cheminformatics

12

Public domain data

ACToRToxNETNTP PubmedIUCLIDOECD Tool Box

Mining tool

Expert

Tox studiesPhys-chempropertiesExposure

Business data

Mining tool

SAR ToolsOASIS DEREKEPA TEST EpiWinEpiSuite ToxTreeOECD Toolbox TopKatECOSAR REACHAcross

Assessment

13

Validated ApproachesValidated Approaches

Tier 2:Bioprofiling

QSAR, e.g., REACHAcross:

Acute mammalian toxicity

EPI Suite:Ecotox &

EnvFate (PBT)

in vitro:Acute mammalian

toxicity & Endocrine activity

BioprofilingDose-response (POD)

Mode of Actionin vitro

Transcriptomics: Chronic, DART, Neuro

BioprofilingDose-response (POD)

Exposure

Calculation of Human equivalent POD

exposure

Screening Level Risk Assessment: Biological Activity Based Margin of ExposureScreening Level Risk Assessment: Biological Activity Based Margin of Exposure

IVIVE: in vitro to in vivo extrapolation

Tier 1: Cheminformatics

Research ApproachesResearch Approaches

Integrated Non-animal Testing Strategy: New and Existing ChemicalsIntegrated Non-animal Testing Strategy: New and Existing Chemicals

Confirm &

Dose-response

Confirm & Dose-response

THE OPPORTUNITY OF BIG DATA & CHEMINFORMATICS

14

Big Data – Rapid Increase in Availability of Chemical and Toxicology Data

15

ECHA Animal & in vitro toxicity dataOver 20,000 compoundsEPA Over 2,000 compoundsEFSA Close to 10,000 compounds

Animal & in vitro toxicity dataAnimal & in vitro toxicity data

Vector (binary descriptor)2D Conformational Substructure Fingerprints

Comparing Chemicals to Establish Structural Similarities

16

Hazard Predicted by Nearest Neighbor

A B90%

17

33383 Unlabeled Chemicals

Similarity Map to300 Labeled Chemicals

Similarity Map to600 Labeled Chemicals

Similarity Map to1387 Labeled Chemicals

Increasing “Similars” with adding new Chemicals

Network Connections → Increasing Information

18

REACH SkinSensitization

REACH allchemicals

Chemical Similarity

19

UL REACHAcross™ Internal Validation

EndpointAcute Toxicity-OralAcute Toxicity-DermalAcute Toxicity-InhalationSkin SensitizationSkin IrritationEye IrritationMutagenicityAcute Aquatic ToxicityChronic Aquatic Toxicity

Total Pos Neg12157 10225 1932

6314 4334 19806184 4812 13724751 2865 1886

15106 13758 134815722 14778 944

3395 600 27952043 1122 9212805 2554 251

Se%777774796983757678

BAC%777674717070737572

BAC%9278

84768475

Sp%777575637156717466

Nine endpoints:REACH 2018 <10 tons/a

Internal validation for unprecedented

number of chemicals

Correctly Identified True

PositivesTrue

Negatives

AverageSe + Sp

AnimalStudies

Accuracy similarto animal test

Cheminformatics (e.g., QSAR) for Green ChemistryCriteria for a useful toxicology tool:

• Assess human health AND environmental end points

• Rapid • Relatively inexpensive • No test material required• User friendly

Resources:USEPA: Toxicity Estimation Software Tool (TEST) https://www.epa.gov/chemical-research/toxicity-estimation-software-tool-testECHA JRC Computational Tools: https://eurl-ecvam.jrc.ec.europa.eu/laboratories-research/predictive_toxicology/qsar_toolsUL REACHAcross: https://www.ulreachacross.com/

The Crossroads of Chemistry and Toxicology: Advancing Greener, Safer Ingredients and Products

Pamela J. Spencer, Ph.D., D.A.B.T.Senior Director of Regulatory Affairs & Product StewardshipANGUS Chemical Company

Toxicology

Chemistry

Overview

§ Our Products

§ Market Challenges

§ New Product Introduction

§ Early safety screening strategies

§ Next steps

§ Q&A

2

OUR PRODUCTS

3

ChemistryANGUS is the only manufacturer in the world that uses propane nitration technology to create a unique set of products.

4

Propane

Nitroalkanes(Nitromethane, Nitroethane)

Nitroalcohols

Aminoalcohols(AMP, TRIS AMINO)

Aminoalcohol Derivatives(Zoldine MS-Plus, ALKATERGE)

Nitration of propane yields our Nitroalkane products

Nitroalcohols are formed via the reaction of nitroalkanes and aldehydes

Aminoalcohols are formed via the reduction of nitroalcohols

Aminoalcohols are converted to derivatives via reactions withaldehydes or carboxylic acids

Technology innovator with over 70 years of nitration expertise

Performance Attributes• Improves the quality and performance of a variety of formulations

• Extremely versatile and can be tailored to solve multiple challenges

• Multifunctional attributes may include several of the following:

• Neutralization • Alkaline pH development

• Corrosion Control • Emulsification (oil/water; water/oil)

• Surfactancy • Moisture Displacement

• pH buffering • Dispersancy

• Moisture Scavenging • Chemical scavenging

• Oxygen Scavenging • Free-radical scavenging

• Many of our products are known for their mildness and are suitable for sensitive applications such as personal care, pharmaceuticals and other life science utilities

5

Revenue Split by Industry

Market Overview

Revenue Split by Geography

Global specialty platform serving a diversified set of industries

North America

EMEA

Asia-Pacific

Latin America

Strong, balanced portfolio:• Product• End Market• Geography

Life Science

Paints

Personal Care

MWF

Rubber

BiocidesOthers*

Others*:• Gas Treating• Mineral Slurries• Leather• Adhesives • Industrial Water

Revenue Split by Geography

MARKET CHALLENGES

7

Ingredient Safety – Ingredient Bans

NEW PRODUCT INTRODUCTION

9

Commercial

R & D

Toxicologists work with R& D chemists & the rest of the project management team from the idea generation stage

Iterative Nature of Early R&D

11

Molecular Design

and Synthesis

Performance /Efficacy

MammalianToxicity

andEcotox

EnvironmentalFate

R&D Teams

Integration of New Tools into NPI

Screen

Lit /database

search

Physical /

chemical

properties

in silico

profiling

Flag issues

for early

testing

12

Can

dida

tes

Disc.Stage

1

Disc.Stage

2

Disc.Stage

3

Dev.Stage

1

Dev.Stage

2

Dev.Stage

3

Dev.Stage

4

Dev.Stage

5

Test Flagsin vivo acute

-Tox- Ecotox

in vitro screen

Env. Fate probe

AcuteStudiesFinish Acute

-Tox- Ecotox

Env. Fate

SelectionRisk review

Select chemical

Development decision

Sub-chronic studies

Chronic studies

Registration

Manufacturing startup

Front End Loading in Research can pay huge dividends in Development:

Greener products, more cost effective, less waste

Early Screens: Toxicology “Red Flags”

• CMRs• Carcinogen• Mutagen• Reproductive/Developmental Toxicant

• PBTs• Persistent• Bioaccumulative• Toxic to the Environment

13

Confirm Relevant QSAR Models for Target Chemistry

QSAR Validation Project

• A number of QSAR models and assessed their applicability for the ANGUS nitroparaffin and derivative chemicals with reference to sensitization, mutagenicity, and ready biodegradability

• TopKat QSAR model not recommended for future use for Angus molecules

• In silico models were improved by acquiring additional data to enhance the training sets

• For mutagenicity the QSAR tools appear to have some degree of reliability

• In the case of skin sensitization, it is evident that providing a predictive set of in vitro and QSAR models is complex and needs further research

• conflicting results from in vitro assays compared with in vivo assay may be the result of the potential for in vivo skin metabolism, corrosive properties of these compounds

• different vehicles used in performing the tests

• For biodegradability when multiple models were used some reliability was also found.

14

Flow Chart of Screening Level Assessment & Green Chemical Evaluation Process

15

ExistingData

CandidateChemical

AssessData

Screening Level Assessment (SLA)

Report

Yes

No

Data Gaps

Relevant Data

PredictEffects

O

CH3

CH3

CH3CH3

O

CH3

CH3 CH3

CH3

Comparison Table

Outcome

ExpertJudgment

CompTox

Literature Reports•Name•Synonyms•CAS Number•Structure•Sub-structure•Chemical analogs

read across

Mammalian Toxicity • Oasis™ -Toolbox/Times• EPA Chemistry Dashboard™

EcoToxicity& Env. Fate • Oasis™ -Times/Catabol• EpiWin™

ExpertJudgment

Expert Judgment

EARLY SAFETY SCREENING STRATEGIES IN PRACTICE

16

It all starts with a structure!

Which one is safer?17

Step 1. Read Across to a Chemically Similar Analog*

Endpoint Analog

Acute Oral LD50 688 mg/kg Rats

Acute Dermal LD50 ~2000 mg/kg Rats

Skin Irritation/Corrosion Corrosive

Eye Irritation/Corrosion Corrosive

Skin Sensitization Sensitizer LLNA

Repeated Dose Mild Liver effects at 1000 mg/kg

Reproduction NOAEL 1000 mg/kg

Genotox Negative in all in vitro and in vivo studies

Biodegradation 57% degradation in 56 days in OECD 301F

Acute Aquatic Toxicity EC/LC50 >10-100

*The chemical structures and following data are meant to serve as an illustrative example and do not represent an actual data for the structures.

New chemical candidate Read across analog

Step 2. OECD Toolbox Modeling for Candidate Chemical

Endpoint Model(s) Result

Acute Oral Cramer Original/Extension

Low Toxicity

Skin Irritation/Corrosion Inclusion/Exclusion rules by BfR

Inclusion Rules Not Met

Eye Irritation/Corrosion Inclusion/Exclusion rules by BfR

Inclusion Rules Not Met

Skin Sensitization OASIS v.1.4 No Alerts

Genotox alerts by ISS No Alerts

Repeated Dose HESS Not Categorized

Developmental/Reproduction DART Scheme v.1.0. No Alerts

Primary Biodegradation Biowin 4 Degradation in weeks to months

Ecotoxicity Verharr (Modified) Determination Not Possible

Report Out Weight of Evidence Safety Assessment

• No significant alerts resulted from read across or modeling

• OECD ToolboxØModeling results were supported by the testing for chemical

analog, except for skin sensitization

ØThree endpoints (skin/eye irritation/corrosion and ecotoxicity) where the target molecule was out of scope for the models utilized

20

Recommendations/Next Steps

• Advanced candidate chemical to Step 3: in vitro testing to better characterize endpoints out of scope of modeling tools and/or did not align with chemical analog data

• Invitro Test Results:ØSkin/eye irritation – non irritating

ØSkin sensitization – not a skin sensitizer

• Biodegradation/ecotoxicity screen may also be necessary• Continue engagement with Project Management Team for NPI

to plan for key studies need to ultimately register new product in target global regions

21

Future Work

• Collaborative project underway with UL to validate REACHAcross ™ Tool for use with ANGUS Chemistry

• Project will be modeled will utilize data set from previous QSAR validation work

• Goal – determine if REACHAcross™ tool has improved domain coverage over other QSAR tools currently utilized

22

Summary• Concerns over product safety can determine success or failure

of product

• Companies can now apply quick and inexpensive tools for estimating EH&S issues early in product development (ID Red Flags)

• When a concern is highlighted via these tools a thoughtful approach to resolving the concern with toxicology data can be prepared in advance to avoid adverse findings in screening analysis

• Increases the ability to launch sustainable new products more cost effectively

23

Thank You