Virtual ICH M7 Expert Review Workshop Dr. Sian Ives Sales Manager - - PowerPoint PPT Presentation

virtual ich m7 expert review workshop
SMART_READER_LITE
LIVE PREVIEW

Virtual ICH M7 Expert Review Workshop Dr. Sian Ives Sales Manager - - PowerPoint PPT Presentation

Virtual ICH M7 Expert Review Workshop Dr. Sian Ives Sales Manager sian.ives@lhasalimited.org ICH M7 Assessment and Control of DNA Reactive (Mutagenic) Impurities in Pharmaceuticals to Limit Potential Carcinogenic Risk Identification


slide-1
SLIDE 1

Virtual ICH M7 Expert Review Workshop

Sales Manager

sian.ives@lhasalimited.org

  • Dr. Sian Ives
slide-2
SLIDE 2

ICH M7

“Assessment and Control of DNA Reactive (Mutagenic) Impurities in Pharmaceuticals to Limit Potential Carcinogenic Risk”

Identification Categorisation Qualification Control

slide-3
SLIDE 3

ICH M7 section 6

  • Hazard assessment involves an initial analysis of actual and

potential impurities by conducting database/literature searches

  • You may use the Ames (in vitro) assay to characterise risk/hazard
  • Or use in silico mutagenicity predictions in its place
  • If you submit in silico predictions, you will need:
  • Two predictions: one expert rule-based and one statistical-based
  • “The absence of structural alerts from both is sufficient to

conclude that the impurity is of no mutagenic concern”

  • Expert review
slide-4
SLIDE 4
slide-5
SLIDE 5

In silico workflow under M7

Evaluate drug substance, impurities, degradants, intermediates Databases, in-house, literature Known mutagen Limit according to TTC or present purge argument for loss Treat as non- mutagenic Known non-mutagen

slide-6
SLIDE 6

What is Vitic?

  • Structure searchable toxicity database and chemical

database management system

  • Public toxicity data
  • You can add your own private, in-house data
slide-7
SLIDE 7

Vitic 2018.1 database

  • 20,580 substances
  • Key endpoint record count:
  • Overall call

10,930 substances 39,435 records

  • Genetox in-vitro summary 14,234 substances

17,216 records

  • Genetox in-vitro

13,657 substances 221,958 records

  • Genetox in-vivo

3,037 substances 12,671 records

  • Carcinogenicity

3,870 substances 16,464 records

slide-8
SLIDE 8
slide-9
SLIDE 9
slide-10
SLIDE 10

In silico workflow under M7

In silico predictions expert / statistical Evaluate drug substance, impurities, degradants, intermediates Databases, in-house, literature Known mutagen Limit according to TTC or present purge argument for loss Treat as non- mutagenic Known non-mutagen

slide-11
SLIDE 11

What is Derek Nexus?

  • Expert system for prediction of toxicology
  • Comprises of 132 Alerts for mutagenicity
  • Written by expert scientists from detailed expert analysis
slide-12
SLIDE 12
  • Alert screenshot
slide-13
SLIDE 13

Skin permeability Presence of toxicophore Molecular weight Species Metabolic activation

How does Derek Nexus predict?

slide-14
SLIDE 14
slide-15
SLIDE 15

What is Sarah Nexus?

  • Statistical model for prediction of mutagenicity
  • Model uses a dataset of ~10K compounds
  • Uses a Self Organising Hypothesis Network (SOHN) methodology*

*Thierry Hanser, Chris Barber, Edward Rosser, Jonathan D Vessey, Samuel J Webb and Stéphane Werner: Self organising hypothesis networks: a new approach for representing and structuring SAR knowledge. Journal of Cheminformatics. 2014, 6:21.

slide-16
SLIDE 16

How does Sarah Nexus predict?

  • User enters query compound
  • Query compound is fragmented
  • Sarah uses the dataset to identify relevant hypotheses
  • A confidence value is calculated
  • Considering Ames Result and Similarity of supporting examples

… … Hypothesis 1: Hypothesis 2: POSITIVE 72% Confidence

Decreasing similarity

Supporting Examples: Supporting Examples:

Positive 65% confidence Positive 78% confidence

slide-17
SLIDE 17
slide-18
SLIDE 18

ICH M7 classification tool

slide-19
SLIDE 19

In silico workflow under M7

Expert Review In silico predictions expert / statistical Evaluate drug substance, impurities, degradants, intermediates Databases, in-house, literature Known mutagen Both predict positive Both predict negative Ames test Limit according to TTC or present purge argument for loss Treat as non- mutagenic Known non-mutagen Disagree / fail to predict

slide-20
SLIDE 20

Virtual ICH M7 Expert Review Workshop

Senior Account Manager

accountmanagement@lhasalimited.org

  • Dr. Dave J. Yeo
slide-21
SLIDE 21

Introduction

  • Example compounds with conflicting expert rule-based

and statistical-based results

  • Derek and Sarah Results:
  • Will be presented
  • You will be asked to assess each example and vote

accordingly

  • Class 3 – Alerting structure. Control to TTC or test…
  • Class 5 – Treat as a non mutagen…
  • Unsure…
  • Lhasa will then present an expert call.*

*All expert calls made on the examples are solely the opinion of experts at Lhasa Limited and are made here for guidance purposes only

slide-22
SLIDE 22

Example 1

slide-23
SLIDE 23

Review high level predictions

?

Expert Review M7 classification

slide-24
SLIDE 24

Review the expert prediction

?

Expert Review M7 classification

  • Alert 346: Trichloromethyl aromatic compound
  • Any aromatic carbon linked to a trichloromethyl group.
slide-25
SLIDE 25

Review the statistical prediction

?

Expert Review M7 classification

  • 50% Positive in Sarah Nexus
  • 5 most relevant nearest neighbours are all positive
  • Negative nearest neighbours are dissimilar (not trichloromethyl aromatic compounds).
slide-26
SLIDE 26

Expert review

?

Expert Review M7 classification

  • Predicted positive
  • Fires alert 346: Trichloromethyl aromatic compound.
  • Predicted positive - 50% confidence
  • 5 most relevant nearest neighbours are all positive

and fire Derek alert 346.

slide-27
SLIDE 27

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-28
SLIDE 28

M7 Classification

?

Expert Review M7 classification

Classification

  • Predicted positive by Derek Nexus (alert 346)
  • No reason to doubt alert
  • Predicted positive by Sarah Nexus
  • 100% confidence-exact match
  • 5 most relevant nearest neighbours are all positive

and fire Derek alert 346.

slide-29
SLIDE 29

M7 Classification

?

Expert Review M7 classification

Class 3

Alerting structure

slide-30
SLIDE 30

Example 2

slide-31
SLIDE 31

Review high level predictions

?

Expert Review M7 classification

slide-32
SLIDE 32

Review the expert system

  • No alerts fired
  • Potential aromatic amine present in the molecule.

?

Expert Review M7 classification

slide-33
SLIDE 33

Review the statistical prediction

  • Two negative hypotheses.

?

Expert Review M7 classification

slide-34
SLIDE 34

Review the statistical prediction

  • Parent amine present in similar Ames negative compounds.

?

Expert Review M7 classification

slide-35
SLIDE 35

Expert Review

  • Inactive prediction
  • No misclassified or unclassified features.
  • Negative prediction
  • Compounds in the training set are similar and Ames

negative

  • Parent amine is Ames negative.

?

Expert Review M7 classification

slide-36
SLIDE 36

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-37
SLIDE 37

M7 Classification – Example 2

  • Predicted inactive by Derek Nexus
  • No reason to overturn the prediction.
  • Predicted negative by Sarah Nexus
  • Closely related examples are Ames Negative.

?

Expert Review M7 classification

Classification

slide-38
SLIDE 38

M7 Classification – Example 2

?

Expert Review M7 classification

Class 5

No alerts or alerting with sufficient data to demonstrate lack of mutagenicity

slide-39
SLIDE 39

Example 3

Compound

slide-40
SLIDE 40

Review high level predictions

?

Expert Review M7 classification

Compound Nexus - ICH M7 Summary result

slide-41
SLIDE 41

Review the expert prediction

?

Expert Review M7 classification

  • Alert 318: Azide, hydrazoic acid or azide salt
  • Any atom linked to an azide group.
slide-42
SLIDE 42

Review the statistical prediction

?

Expert Review M7 classification

An outside domain outcome was obtained as the current training set does not include any example covering the sulfonyl azide fragment.

slide-43
SLIDE 43

Review the statistical prediction

?

Expert Review M7 classification

By replacing the outside domain sulfonyl azide fragment by an azidomethyl fragment, a positive prediction was reported matching the existing azide hypothesis.

slide-44
SLIDE 44

Additional information (from Vitic)

?

Expert Review M7 classification

No Ames data available for N-(4-Azidosulfonylphenyl)acetamide (OOD)

  • r for N‐(4‐(Azidomethyl)phenyl)acetamide (predicted positive).

For the 1st nearest neighbour to N‐(4‐(Azidomethyl)phenyl)acetamide: Benzyl azide 622-79-7 (most relevant example), positive results observed in TA100+S9. Overall calls from various datasets for 622-79-7: Positive Ames data were reported for various benzyl azide derivatives.

slide-45
SLIDE 45

Expert review

?

Expert Review M7 classification

  • Active prediction
  • Fires alert 318 for Azide, hydrazoic acid or azide salt
  • No data available to overturn the prediction.
  • Out of domain sulfonyl azide fragment
  • Sarah Nexus does have a positive hypothesis for azides.
slide-46
SLIDE 46

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-47
SLIDE 47

M7 Classification – Example 3

?

Expert Review M7 classification

Classification

  • Positive prediction from Derek Nexus (alert 318)
  • No reason to doubt.
  • Out of domain in Sarah due to sulfonylazide fragment
  • Sarah Nexus has a positive hypothesis for azides.
slide-48
SLIDE 48

M7 Classification – Example 3

?

Expert Review M7 classification

Class 3

Alerting structure

slide-49
SLIDE 49

Example 4

slide-50
SLIDE 50

Review high level predictions

?

Expert Review M7 classification

slide-51
SLIDE 51

Review the expert prediction

?

Expert Review M7 classification

  • Alert 027: Alkylating agent
  • No reason to doubt predicted activity.
slide-52
SLIDE 52

Review the statistical prediction

?

Expert Review M7 classification

  • No hypothesis is identified
  • The relevant nearest neighbours are all positive and fire Derek alert 027
  • Negative nearest neighbours are dissimilar and are not alkylating agents.
slide-53
SLIDE 53

Expert review

?

Expert Review M7 classification

  • Positive prediction
  • No reason to doubt.
  • Equivocal prediction
  • No hypothesis identified
  • Relevant nearest neighbours in the training set are positive
  • Nearest neighbours that are negative are dissimilar and are not

alkylating agents.

slide-54
SLIDE 54

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-55
SLIDE 55

M7 Classification – 5689-83-8

?

Expert Review M7 classification

Classification

  • Derek Nexus
  • No reason to doubt Derek Nexus positive

prediction (alert 027)

  • Sarah Nexus
  • Equivocal prediction overruled as relevant

examples are positive

slide-56
SLIDE 56

M7 Classification – 5689-83-8

?

Expert Review M7 classification

Class 3

Alerting structure

slide-57
SLIDE 57

Example 5

slide-58
SLIDE 58

Review high level predictions

?

Expert Review M7 classification

slide-59
SLIDE 59

Review the expert prediction

?

Expert Review M7 classification

  • Negative prediction
  • No misclassified or unclassified features.
slide-60
SLIDE 60

Review the statistical prediction

?

Expert Review M7 classification No hypothesis is identified. Activity of nearest neighbours is associated with their ability to act as alkylating agents or the epoxide, neither are relevant to the query.

slide-61
SLIDE 61

Review the statistical prediction

?

Expert Review M7 classification No hypothesis is identified. Parent alcohol and sulfonic acid are both negative examples in the training set.

slide-62
SLIDE 62

Expert review

?

Expert Review M7 classification

  • Negative prediction
  • No misclassified or unclassified features.
  • Equivocal prediction
  • No hypothesis identified
  • Activity of nearest neighbours is associated with structural features

enabling them as alkylating agents

  • Parent alcohol and sulfonic acid are negative examples in the

training set.

slide-63
SLIDE 63

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-64
SLIDE 64

M7 Classification – Example 5

  • Derek Nexus
  • Inactive prediction with no misclassified or

unclassified features.

  • Sarah Nexus
  • Equivocal prediction overruled as positive examples

are not relevant and parent compounds are negative examples in the training set.

?

Expert Review M7 classification

Compound

slide-65
SLIDE 65

M7 Classification – Example 5

?

Expert Review M7 classification

Class 5

Class 5 - No alerts or alerting with sufficient data to demonstrate lack of mutagenicity

Compound

slide-66
SLIDE 66

Example 6

slide-67
SLIDE 67

Review high level predictions

?

Expert Review M7 classification

slide-68
SLIDE 68

Review the expert prediction

?

Expert Review M7 classification Alert 315: Acid halide Sulphonyl chloride functional group match sulphonyl halide toxicophore III However risk of false positives flagged in Ames tests in DMSO. [Amberg A et al., Organic Process Research and Development (2015), 19(11), 1495-1506] 5-strain Ames negative results for Benzene sulphonyl chloride (98-09-9) in acetonitrile.

slide-69
SLIDE 69

Additional information (from literature)

?

Expert Review M7 classification

Amberg et al., Organic Process Research and Development (2015), 19(11), 1495-1506

slide-70
SLIDE 70

Additional information (From Vitic)

?

Expert Review M7 classification

No Ames data available on 15945-07-0 2,4,5-Trichlorobenzenesulfonyl chloride 5-strain Ames negative results for Benzene sulphonyl chloride (98-09-9) in acetonitrile. [Amberg A et al., Organic Process Research and Development (2015), 19(11), 1495-1506] (false positives due to reactivity with DMSO discarded).

slide-71
SLIDE 71

Review the statistical prediction

?

Expert Review M7 classification Compounds 1 and 7 confirmed absence of activity of benzene with chloro substituents.

Compounds from training set are dissimilar to the test compound. i.e., none of the examples contain a sulphonyl chloride group.

slide-72
SLIDE 72

Expert review

?

Expert Review M7 classification

  • Predicted positive
  • Fires alert 315 for acid halide, however Ames

positive calls challenged as potential for false positives due to reactivity with DMSO

  • 5-strain Ames negative results for similar

compound benzene sulphonyl chloride (98-09-9) in acetonitrile.

  • Inactive prediction
  • Nearest neighbours from Sarah training dissimilar to

the test compound i.e. no sulfonyl chloride group.

slide-73
SLIDE 73

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-74
SLIDE 74

M7 Classification – Example 6

?

Expert Review M7 classification

Classification

  • Derek positive prediction (alert 315)
  • Compound may react with DMSO to produce a reactive,

mutagenic intermediate

  • The alert fires to explain artefactual activity that may be seen
  • Similar compounds Ames negative in other solvents.
  • Sarah negative prediction
  • Nearest neighbours from Sarah training dissimilar to the test

compound i.e. no sulfonyl chloride group

  • Similar compound in the training set, benzene sulphonyl

chloride (98-09-9) is negative.

  • Overrule Derek prediction
slide-75
SLIDE 75

M7 Classification – Example 6

?

Expert Review M7 classification

Class 5

No alerts or alerting with sufficient data to demonstrate lack of mutagenicity

slide-76
SLIDE 76

Example 7

slide-77
SLIDE 77

Review high level predictions

?

Expert Review M7 classification

slide-78
SLIDE 78

Review the expert prediction

?

Expert Review M7 classification

slide-79
SLIDE 79

Review the expert prediction

?

Expert Review M7 classification

slide-80
SLIDE 80

Review the statistical prediction

?

Expert Review M7 classification

Most training set examples have an aromatic group attached to the alkene (C3 position) which Derek Nexus has described as having decreased mutagenic activity.

slide-81
SLIDE 81

Review the statistical prediction

?

Expert Review M7 classification

  • There is no protocol information.
slide-82
SLIDE 82

Expert review

?

Expert Review M7 classification

  • Plausible prediction
  • Fires alert 664 for alpha,beta-unsaturated compounds
  • Mutagenicity related to electronegativity and steric hindrance

around the double bond

  • Aryl substituents at C3 are typically negative or reduced

mutagenicity.

  • Negative prediction
  • Negative compounds mostly have an aryl substituent at C3

position

  • Most similar compound, 30418-35-2 is negative however there is

no protocol information.

slide-83
SLIDE 83

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-84
SLIDE 84

M7 Classification – Example 7

?

Expert Review M7 classification

  • Predicted positive
  • No data to support overturning the prediction.
  • Predicted inactive
  • Many of the supporting examples are not relevant.
slide-85
SLIDE 85

M7 Classification – Example 7

?

Expert Review M7 classification

Class 3

Alerting structure

slide-86
SLIDE 86

Example 8

Compound

slide-87
SLIDE 87

Review high level predictions

?

Expert Review M7 classification

Compound

slide-88
SLIDE 88

Review the expert prediction

?

Expert Review M7 classification

slide-89
SLIDE 89

Review the expert prediction

?

Expert Review M7 classification

  • Candidate for alert 351
  • Phenyl substitution of secondary amine excluded and electron withdrawing group

present on the aromatic ring deactivating

  • Primary amine group does not activate due to ortho aromatic amine substitution

which may cause steric hindrance.

slide-90
SLIDE 90

Review the statistical prediction

?

Expert Review M7 classification

351

slide-91
SLIDE 91

Review the statistical prediction

?

Expert Review M7 classification Good strain data for both negatives

slide-92
SLIDE 92

Expert review

?

Expert Review M7 classification

  • Negative with no misclassified or unclassified features
  • Near miss of alert 351
  • Exclusion based on limitations in substitution allowed on second

amine group in ortho diamines

  • Exclusion based on available evidence rather than negative data.
  • Sarah Nexus identifies a positive hypothesis for aromatic amine

which is overruled by nearest neighbours

  • Two negative nearest neighbours are relevant and electronically

similar to both amines present in query structure

  • Data lacking for ortho-substituted aromatic amine and there is a

data gap for this sub-class.

slide-93
SLIDE 93

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-94
SLIDE 94

M7 Classification – Example 8

  • Derek inactive prediction
  • Exclusion from alert 351 based on steric hindrance

around aromatic amine

  • Query compound may adopt a conformation with less

steric hindrance and potential to intercalate.

  • Sarah negative prediction
  • Two most similar relevant examples negative
  • No data for this type of ortho substitution relating to

steric influence.

  • While there is limited data available relating to this type of
  • rtho substitution of an aromatic amine, the same

substitution in the para position gave negative results. No need to overturn Derek Nexus and Sarah Nexus.

?

Expert Review M7 classification

Compound

slide-95
SLIDE 95

M7 Classification – Example 8

?

Expert Review M7 classification

Class 5

No alerts or alerting with sufficient data to demonstrate lack of mutagenicity Compound

slide-96
SLIDE 96

Example 9

slide-97
SLIDE 97

Review high level predictions

?

Expert Review M7 classification

slide-98
SLIDE 98

Review the expert prediction

?

Expert Review M7 classification

slide-99
SLIDE 99

Review the statistical prediction

?

Expert Review M7 classification

  • Equivocal prediction
  • Training set examples are not relevant to query
  • Check additional information…
slide-100
SLIDE 100

Review the statistical prediction

?

Expert Review M7 classification

  • Close match with training set compound
  • Ames positive with results in two strains
  • No reason to doubt the positive.
slide-101
SLIDE 101

Expert review

?

Expert Review M7 classification

  • Inactive prediction, however the compound contains misclassified

features.

  • Compound is in Lhasa reference Ames test dataset as a known

false negative.

  • Equivocal prediction
  • Additional information shows a closely related training set example
  • This example is Ames positive and no reason to doubt data.
slide-102
SLIDE 102

Please make your selection

  • 1. Class 3 – Alerting structure
  • 2. Class 5 – No alerts or alerting with sufficient data

to demonstrate lack of mutagenicity

  • 3. Unsure
slide-103
SLIDE 103

M7 Classification – Example 9

?

Expert Review M7 classification

Class 3

Alerting structure

slide-104
SLIDE 104

Thank You