Structure-Activity Relationships in Toxicology:

Introduction (and a case study)Part I.

Romualdo BenigniRomualdo BenigniIstituto Superiore di Sanita’ RomeIstituto Superiore di Sanita’ Rome

Three Rs:

Replacement, Reduction, and Refinement of animal testing

with the aims of shortening times of toxicity testing, protecting

animal health and welfare, and saving money

Traditional toxicology has been the major source of information

Now, opportunities to accept “alternative” approaches

Traditional toxicology has been the major source of information in EU

Now, opportunities to accept “alternative” approaches

 

REACH: EU new regulation of chemicals

Registration, Evaluation and Assessment of Chemicals

•non-testing methods, including (Q)SARs, read-across and chemical category approaches will be used more extensively and more systematically than under previous EU legislation….

 

Read-Across:data gaps filling approach; information for one or more source chemicals is used to make a prediction for a target chemical, considered to be similar in some way.

(Q)SAR:(Quantitative) Structure-Activity Relationships

Chemical category:group of chemicals whose physicochemical and toxicological properties are likely to be similar or follow a regular pattern as a result of structural similarity

OECD Principles

To facilitate the consideration of a (Q)SAR model for regulatory purposes, it should be associated with the following information:

• 1) a defined endpoint

• 2) an unambiguous algorithm

• 3) a defined domain of applicability

• 4) appropriate measures of goodness-of–fit, robustness and predictivity

• 5) a mechanistic interpretation, if possible

A case study: (Q)SARs for mutagens and carcinogens

• Theory more advanced than in other toxicological fields (e.g., knowledge on action mechanisms)

• Suitable to show the different approaches to (Q)SAR

Chemical mutagens and carcinogens:background information

Mechanistic findings at the basis of the science and regulation of mutagens and carcinogens

• Millers’ electrophilic (DNA-) reactivity theory of carcinogenesis (not including nongenotoxic carcinogens)

• Chemical mutagenicity: Malling’s in vitro metabolic activation (S30, S9); Salmonella, or Ames’ test for DNA-reactive chemicals

• Structure-Activity (carcinogenicity) Relationships (Ashby’s Structural Alerts)

Mechanistic findings at the basis of the science and regulation of mutagens and carcinogens

• Because of the success of Millers’ electrophilic reactivity theory of carcinogenesis, and of Ames’ test:

major research efforts on the hypothesis Mutation = Cancer

• Later on, recognition of nongenotoxic carcinogens

Looking for further mutagenicity Short-Term Tests (STT) to predict carcinogenicity

• Hypothesis to cover the spectrum of cancer-relevant factors, use: different genetic endpoints (gene mutation, chromosomal damage), different cells (bacterial, mammalian), and animals (ADME)

• Development of > 100 STTs based on:

mutagenicity (e.g., gene mutation in mammalian cells, chromosomal aberrations, aneuploidy)

other genotoxic events (e.g., DNA damage)

STTs to predict carcinogenicity: state-of-the-art

• Mutagenicity = Carcinogenicity ? Only within a limited area of the chemical space, i.e., DNA-reactive

chemicals

• DNA-reactive chemicals induce cancer, and a wide spectrum of mutations

• Most predictive mutagenicity-based assay: Ames test

• Other in vitro assays (e.g., clastogenicity), when Ames-negative : no correlation with carcinogenicity

• No reliable in vivo STTs (e.g., micronucleus) available

Benigni R. et al., Exp.Opinion Drug Metab.Toxicol., 2010, 6: 1-11. Zeiger E Regulat.Pharmacol.Toxicol. 1998;28:85-95.

Ames test neg pos

Carcinogenicity

233 76 Non DNA-reactive 136 34

DNA-reactive 79 277

Results from 835 chemicals in ISSCAN v3a http://www.iss.it/ampp/dati/cont.php?id=233&lang=1&tipo=7

Ames test versus rodent carcinogenicity

Ames identifies DNA-reactive carcinogens

Non carcinogens

Carcinogens

Ames test neg pos

Carcinogenicity

233 76 Non DNA-reactive 136 34

DNA-reactive 79 277

Results from 835 chemicals in ISSCAN v3a http://www.iss.it/ampp/dati/cont.php?

id=233&lang=1&tipo=7

Ames test versus rodent carcinogenicity

Ames mutagen: 80% probability of being a carcinogen

Non carcinogens

Carcinogens

Backing up the mutagenicity STTs with Structure-Activity concepts

•To model / predict the Ames test

•To complement the Ames test

Structure-activity relationship concepts:

application to different issues, through different approaches

Coarse-grain Structure Alerts

Fine-tuned Quantitative Structure-Activity Relationships (QSAR)

Structure Alerts (SA)

Functional groups or Substructures

linked to

toxic (carcinogenic / mutagenic) effects of chemicals

Several Azo-dyes are carcinogenic

Butter yellow

• Food colorant

• Carcinogenic

NH2

N

N

(generation of aromatic amines)

Hydrophilic sulfonic acid groups generate water-soluble compounds

C.I. Food Black 1 (E 151)

• Safe colorant

• All split products are strongly hydrophilic C

NH

NN

N

N

OH

O- O

-

O-

O-

O

O

O O

O

OOO

O

S S

S

S

CH3

SA: Aromatic Diazo

Modulating factor: Hydrophilic sulfonic acid

The modulating factors diminish or abolish the SA effect

CNH

NN

N

N

OH

O- O

-

O-

O-

O

O

O O

O

OOO

O

S S

S

S

CH3

C.I. Food Black 1 (E 151)

Mechanistic findings at the basis of the science and regulation of mutagens and carcinogens

• Millers’ electrophilic reactivity theory of carcinogenesis

• Ames’ test, in vitro model of the carcinogenicity mechanisms

• Ashby’s theoretical model of carcinogenicity (compilation of Structural Alerts)

2

O

NN

N N

CH3

CH3

CH3 N

N

O

CH2 CH CH2

CH2

O

CO NH2

O

NH2

CH2Cl

CH

NH

CH CH2 CH

CH

CH2 CH

CH2

N Cl

NCH2

CH2

Cl

CHCH2

OC

O

CH CH Cl

N

CH2

OH

CH2

CH

CH ONH

N

CH3

CH3

CH2

S OO

OCH3

NO

CH2

CH2

Cl

Ashby’s

Poly-carcinogen

Ashby (1995) Environ.Mutag. 7: 919-921

Some alerts accompanied by detoxifying (modulating) factors

2

O

NN

N N

CH3

CH3

CH3 N

N

O

CH2 CH CH2

CH2

O

CO NH2

O

NH2

CH2Cl

CH

NH

CH CH2 CH

CH

CH2 CH

CH2

N Cl

NCH2

CH2

Cl

CHCH2

OC

O

CH CH Cl

N

CH2

OH

CH2

CH

CH ONH

N

CH3

CH3

CH2

S OO

OCH3

NO

CH2

CH2

Cl

J Ashby and Tennant R W (1988) Mutat Res 204:17-115

Further elaboration for, e.g., :

• Aromatic amino, substituted amino

• Aromatic nitro

• Aromatic –NR2

• Chlorinated olefins

• PAH

etc…

Ashby J (1978) Cancer 37:904-923

SA: chemical class that provokes toxic effects through one or few

shared mechanisms of action

direct-acting carcinogens: e.g., epoxides, aziridines, sulfur and

nitrogen mustards, α-haloethers, and lactones

C C

O

C C+

O- DNA

Strained ring Carbonium ion Alkylation

Metabolically activated carcinogens: e.g., aromatic amines

Ac

N

Ac

OAc

N

H

O SO3

N

Ac

O

N

H

Ac

N

OH

Ac

Acetyl CoA

Cytochrome P450s

Trans-acetylases

Electrophilic metabolites

Covalent binding to DNA

Toxic effect (mutation and/or cancer)

NH

OH

NH2

SA: chemical class that provokes toxic effects through one or few

shared mechanisms of action

SA: chemical class that provokes toxic effects through one or few

shared mechanisms of action

complex carcinogens: e.g., aliphatic halogens

From genotoxic to epigenetic, with increasing degree of halogenation and depending on the carbon skeleton

Short-chain monohalogenated alkanes (and alkenes) direct-acting alkylating agents;

Dihalogenated alkanes: alkylating or cross-linking agents.

Polyhaloalkanes: by free radical or nongenotoxic mechanisms, or reductive dehalogenation to yield haloalkenes.

Halogenated cycloalkanes (and cycloalkenes): possibly epigenetic or direct alkylation after metabolic transformation

Toxtree: Rulebase for mutagens / carcinogens

Structure-based approach consisting of:

-New compilation of Structure Alerts (genotox (DNA-reactive) and non-genotox)

-Three mechanistically-based QSARs for congeneric classes (aromatic amines, aldehydes)

Expert system Toxtree (version 1.6)

Open-source, freely available: http://ecb.jrc.it/qsar/qsar-tools/index.php?c=TOXTREE

Primary aromatic amine

Frequency (%)

0 5 10 15 20 25 30 35

Various alkylating

Epoxides and aziridines

Aliphatic halogens

Simple aldehyde

Quinones

Hydrazines

Aliphatic azo and azoxy

Isocyanate and isothiocyanate groups

Alkyl carbamate and thiocarbamate

Polycyclic Aromatic Hydrocarbons

Alkyl and aryl N-nitroso groups

Aliphatic N-nitro group

Aromatic nitroso

Nitro-aromatic

Aromatic amines

Coumarins and Furocoumarins

Non genotoxic SAs

NoSAs

Profile of the Kirkland’s database on carcinogens

Toxtree 2.5 update:

more rulebases

Many Toxtree rulebases

included in the

OECD (Q)SAR Toolbox

Checking the agreement between

SAs and experimental results

A chemical with a SA (with no modulating factors)

is predicted as potentially toxic

ROC graph

False Positive rate

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Tru

e P

ositi

ve R

ate

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Perfect

Random

Horrible

Perfect

Random

Horrible

ROC graph: A simple, graphical way of comparing predictions with results

True positive rate = (Positives predicted as positive) / (Real positives)

= Sensitivity

False Positive Rate = (Negatives predicted as positive) / (Real negatives)

= 1 - Specificity

False positive rate

True positive rate

Toxtree SAs: agreement with Carcinogenicity and Salmonella (Ames)

ISSCAN v3a database

SAs identify Ames-mutagens with 80% accuracy, comparable to inter-laboratory variability (80-85%)

Perfect

Random

All wrongs

False Positive Rate

0.0 0.2 0.4 0.6 0.8 1.0

Tru

e P

ositi

ve R

ate

0.0

0.2

0.4

0.6

0.8

1.0

STY

SA_BB & STY

SA_BB

Carcinogenicity prediction: Salmonella (Ames) versus SAs

SAL + SAs

SAs

SAL

ISSCAN v3a database

Salmonella and SAs identify genotoxic

(DNA-reactive) carcinogens with similar accuracy,

and are not complementary to each other

Knowledge on DNA-reactivity (coded in SAs):

• Reliable enough to predict Salmonella results, and identify many carcinogens

• Identify human carcinogens

• Basis for successful priority setting in NTP bioassays (70% carcinogens among structurally suspect chemicals, only 10% among high exposure chemicals)

• Contribution to reduce DNA-reactive carcinogens among synthetic chemicals (pesticides, pharmaceuticals)

Success story: priority setting by human experts

Out of 400 chemicals tested by NCI/NTP:

• 2/3 selected as suspect carcinogens (n=267)

68% carcinogenic (n=187)

• 1/3 selected on production/exposure considerations (n=133)

20% carcinogenic (n=26); 6.8% positive in two species (n=9)

Fung, Barrett, and Huff (1995) Environ.Health Perspect. 103: 680-683

Which use for the Structure Alerts ?

Which use for the Structure Alerts ?

Great tool for coarse-grain characterization of the chemicals:

• Description of sets of chemicals

• Preliminary hazard characterization

• Category formation (e.g., regulation, fine-tuned QSAR, etc…)

• Priority setting (enriching the target)