An Overview of Recent Skin Sensitisation Work at Lhasa Overview of Rec… · Freund’s complete adjuvant test. Freund’s complete adjuvant test (modified) Split adjuvant test. Single

An Overview of Recent Skin Sensitisation Work at Lhasa

Senior Scientist

[email protected]

Donna Macmillan

Agenda

• Skin sensitisation and legislation

• Negative predictions for skin sensitisation

• Lhasa’s defined approach using in silico / in chemico / in

vitro data

• Conclusions

Skin sensitisation and legislation

What is skin sensitisation?

• Common occupational disease

• Estimated to cost the EU €600 million and 3 million lostworking days

• High concentrations and/or repeated exposure increasesthe likelihood of becoming sensitised to a specificchemical

• Not life-threatening but is life-long

What is skin sensitisation?

(2) Elicitation:subsequent contact with the samehapten leads to the hapten-proteincomplex triggering the allergen specificT-cells which induces inflammatorycytokines and skin sensitisation

(1) Induction:allergen (hapten) forms a stableconjugate (hapten-protein complex)with carrier proteins within the skininitiating a cascade which ends withproliferation of allergen specific T-cells

Figures taken from OECD 2012, The Adverse Outcome Pathway for Skin Sensitisation Initiated by Covalent Binding to Proteins Part 1: Scientific Evidence, Series on Testing and Assessment, No. 168.

• Develops in two stages:

Legislation

Korea REACH

China REACHTurkey KKDIK

Brazil cosmetics ban USA – Amended TSCA

EU REACHCosmetics Regulation

Negative predictions for skin sensitisation

Negative predictions – Methodology

• Trustworthy negative predictions are important as theconsequence of such a prediction being incorrect issevere• False negatives (sensitisers predicted as non-sensitisers)

can have a detrimental impact on human health

• An existing methodology demonstrably works well formutagenicity in Derek, requiring:• A well-developed endpoint

• A reactivity-driven mechanism

• A sufficiently large dataset

R. Williams et al., Regul. Tox. Pharmacol., 2016, 76, 79-86

• A similar methodology for skin sensitisationwas implemented

• An internal skin sensitisation dataset of > 2500chemicals was used

• Dataset consists of human, mouse and guinea pig data• Overall experimental call is derived using a hierarchical

approach

Human Standard animal Non-standard animal Other animal

Non-standard LLNANon-radioactive LLNABuehler (closed patch) testFreund’s complete adjuvant testFreund’s complete adjuvant test (modified)Split adjuvant testSingle injection adjuvant testSingle injection adjuvant test (modified)Maurer optimisation testOptimisation testOpen epicutaneous testClosed epicutaneous test

LLNA GPMT

BgVVBasketter

Draize testDraize test (altered)Mouse ear swelling testpositive data only

Negative predictions – Dataset

Lhasa skin sensdataset

(n > 2500)

Fragment structures

Fragment library(n > 6500)

Classify fragments

Negative predictions – Fragment library

Negative predictions – Outcomes

Query fragment present in Lhasa dataset?

Query fragment contained in known false negatives? Outcome Example

Yes No No misclassified or unclassified features

Yes Yes Misclassified feature

No n/a Unclassified feature

No alertfired

Query chemical

Negative prediction

dataset

• Misclassified and unclassified features occur infrequently and represent

areas of increased uncertainty, which may require further scrutiny

Derek ‘no alert’ Derek with skin negative predictions

How often is each type of prediction correct? How often does eachtype of prediction occur?

Negative predictions – 5-fold CV

• Misclassified and unclassified features occur infrequently and represent

areas of increased uncertainty, which may require further scrutiny

How often is each type of prediction correct?

Derek ‘no alert’ Derek with skin negative predictions

How often does eachtype of prediction occur?

Negative predictions – Member data

Defined approach for skin sensitisation

• 5 in chemico / in vitro assays have OECD guidelines• DPRA, KeratinoSens™, LuSens, h-CLAT, U-SENS™

• Generally agreed that more than one non-animal assay will be required to replace the LLNA or GPMT

Defined approach

Skin sensitisation assay(s)

DPRA

LLNA

h-CLAT / U-SENS™

KeratinoSens™ / LuSens

GPMT / humanOrganism response

Organ response

Cellular response

Molecular Initiating Event Haptenation

T-cell activation

Activation of DC

Stress response

Skin sensitisation

Adverse Outcome Pathway

KE3

KE2

AO

MIE

KE4

Defined approach

Lhasa’s defined approach

• Our hypothesis:

• Use Derek information alongside assay data (grouped into key events in the AOP)

• Apply exclusion criteria to take into account assay limitations and confidence in the Derek prediction

• This ensures the most relevant information source(s) are used for specific chemicals

DPRAh-CLATKeratinoSens™

U-SENS™LuSens

AO MIE KE2 KE3

Summary of exclusion criteriaExclusion criteria Derek MIE KE2 KE3 Comment

Metabolism Prohapten ✓ ✗ ✓ ✓Assays lacking metabolic competency are

deprioritised as they are less likely to predict prohaptens well

logP> 3.5 ✓ ✓ ✓ ✗ Cell-based assays are deprioritised for

chemicals with a logP > 3.5 (KE3) and logP > 5 (KE2) as more lipophilic chemicals may

lack high solubility in these cell-based assays> 5 ✓ ✓ ✗ ✗

Lysine reactive Exclusive ✓ ✓ ✗ ✓

The Nrf2-ARE pathway is associated with cysteine binding - lysine-reactive chemicals

may not be reliably predicted

Likelihood Equivocal ✗ N/AAlerts with a likelihood of equivocal have

less evidence of skin sensitisation potential than other likelihoods (e.g. certain) and are

thus deprioritised

Negative prediction

Misclassified features ✗ N/A Negative predictions with ‘misclassified

features’ or ‘unclassified features’ are deprioritised as these are associated with

higher uncertainty.Unclassified

features ✗ N/A

Making use of Derek

• Metabolism

• Lipophilicity

• Lysine reactivity

• Likelihood level

• Negative predictions

Hazard prediction

NH2

NH2

1st assay

1st assay

1st assay

EquivocalNon-sensitiser with misclassified or unclassified features

2nd assay

3rd assay

2nd assay

2nd assay

2nd assay

CertainProbablePlausible

Non-sensitiserDoubted

Improbable

Derek alert

outcome

Chemical of interest

Use Derek outcome to determine decision tree

branch

Prioritise in chemico/in vitro assays using exclusion criteria

Run in chemico/in vitro assays in order of AOP (MIE → KE2 → KE3) unless de-

prioritised by exclusion criteria

Hazard prediction using ‘2 out of 3’

approach

Exclusion criteria

sensitiser

sensitiser

non-sensitiser

non-sensitiser

sensitiser

non-sensitiser

Blue italics = Derek outcomeRed arrow = positive resultGreen arrow = negative result

Potency prediction

Human data1-2

Mouse data(EC3)

Combine and curate

Combined dataset

ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 > 𝑢𝑢𝑚𝑚𝑢𝑢𝑚𝑚𝑚𝑚

n = 199

n = 672

n = 762

0 1.0

Non-sensitiser

Pote

ncy

Cat

egor

y

Extreme

Strong

0.5

Query compound with predicted

potency category

Tanimoto Similarity

Top 10 Nearest Neighbours

Moderate

Weak

Very weak

Compounds that fire alert

Basketter human potency category

Basketter human potency category name

GHS category

Equivalent EC3 value (%)3

1 extreme 1A < 0.2

2 strong 1A 0.2 – 2

3 moderate 1B 2 – 20

4 weak 1B 20 – 80

5 very weak/non-sensitiser 2 > 80

6 non-sensitiser 2 negative

1. Basketter et al., Dermatitis, 2014, 11-212. Api et al, Dermatitis, 2017, 299-3073. Basketter, 2016,.Altern. Lab. Anim., 431–436

Potency prediction

Potency prediction

modelNH2

NH2

1st assay

1st assay

1st assay


2nd assay

3rd assay

2nd assay

2nd assay

2nd assay



Improbable

Derek alert

outcome

Basketter 5/6 (GHS 2)



branch


Basketter 1 (GHS 1A)

Basketter 2 (GHS 1A)Basketter 3 (GHS 1B)Basketter 4 (GHS 1B)



Potency prediction using k- nearest neighbours

model


approach

Exclusion criteria

sensitiser

sensitiser

non-sensitiser

non-sensitiser

sensitiser

non-sensitiser


Dataset compilation

22

Key Event Assay + - Total Discordant data Final total

MIE DPRA 120 72 192 n/a 192

2KeratinoSens™ 122 62 184

5 195LuSens 45 32 77

3h-CLAT 113 50 163

25 186U-SENS™ 130 39 169

4 LLNA 173 60 233 human call taken preferentially 240

AO Human 83 46 129

Exclusion criteria Chemical property Assay(s) excluded

Metabolism Hapten none

Lipophilicity 4.07 KE3

Lysine reactivity Yes KE2

Derek likelihood Plausible none

Derek negative prediction n/a -

Example 1in vivo

SensitiserBasketter category 1

GHS 1A

CN

OHHO

O

O

O

S

Example 1

Potency prediction

model

1st assay

1st assay

1st assay


2nd assay

3rd assay

2nd assay

2nd assay

2nd assay



Improbable

Derek alert

outcome




branch







model


approach

Exclusion criteria

sensitiser

sensitiser

non-sensitiser

non-sensitiser

sensitiser

non-sensitiser


in vivoSensitiser

Basketter category 1GHS 1A

Defined approachSensitiser

Basketter category 2GHS 1A

CN

OHHO

O

O

O

S

AOP event Assay / model Outcome

AO DX plausible

KE1 DPRA positive

KE2 deprioritised

KE3 deprioritised


Metabolism prehapten none

Lipophilicity 3.25 none

Lysine reactivity no none

Derek likelihood equivocal none

Derek negative prediction n/a -

CH3

CH3

CH3

HO

Example 2in vivo

Non-sensitiserBasketter category 5/6

GHS 2

Example 2

in vivoNon-sensitiser

Basketter category 5/6GHS 2

Defined approachSensitiser

Basketter category 4GHS 1B

Potency prediction

model

1st assay

1st assay

1st assay


2nd assay

3rd assay

2nd assay

2nd assay

2nd assay



Improbable

Derek alert

outcome




branch







model


approach

Exclusion criteria

sensitiser

sensitiser

non-sensitiser

non-sensitiser

sensitiser

non-sensitiser


CH3

CH3

CH3

HO


AO deprioritised

KE1 DPRA positive

KE2 KeratinoSens™ / LuSens negative

KE3 h-CLAT / U-SENS™ positive

H3C

OHO

OO

S


Metabolism n/a -

Lipophilicity 4.99 KE3

Lysine reactivity n/a -

Derek likelihood n/a -

Derek negative prediction No misclassified or unclassified features -

Example 3in vivo

Non-sensitiserBasketter category 5/6

GHS 2

Example 3

in vivoNon-sensitiser


Defined approachNon-sensitiser


Potency prediction

model

1st assay

1st assay

1st assay


2nd assay

3rd assay

2nd assay

2nd assay

2nd assay



Improbable

Derek alert

outcome




branch







model


approach

Exclusion criteria

sensitiser

sensitiser

non-sensitiser

non-sensitiser

sensitiser

non-sensitiser


H3C

OHO

OO

S


AO DX Non-sensitiser

KE1 DPRA positive

KE2 KeratinoSens™ / LuSens negative

KE3 deprioritised

Performance

Results

30

Conclusions

• A decision tree defined approach has been designedusing exclusion criteria based on known limitations of inchemico/in vitro assays and Derek Nexus

• The defined approach correctly predicts:

• The Basketter potency category for 64%

• The GHS classification for 77% of chemicals in theevaluation dataset

• Martyn Chilton

• Everyone at Lhasa Limited

Acknowledgements

Lhasa Limited

Granary Wharf House, 2 Canal Wharf

Leeds, LS11 5PS

Registered Charity (290866)

Company Registration Number 01765239

+44(0)113 394 6020

[email protected]

www.lhasalimited.org

Thank you very much for your attention

Are there any questions?

An Overview of Recent Skin Sensitisation Work at Lhasa Overview of Rec… · Freund’s complete adjuvant test. Freund’s complete adjuvant test (modified) Split adjuvant test. Single

Documents