View
3
Download
0
Category
Preview:
Citation preview
ICH M7 WorkshopICGM, Tokyo, November 2019
CEO
chris.barber@lhasalimited.org
Chris Barber
Expert review under ICH M7 guidelines
Applying the ICH M7 guidelines
Is the impurity mutagenic? Is the impurity present in the final API?
What are the potential impurities?
Implement a control strategy
reagents,
intermediates, side
reactions, degradants…
Ames test or in
silico methods
Analytical or in
silico methods
Applying the ICH M7 guidelines
Database
searching
Expert prediction
Statistical
prediction
Expert
assessment
Classification
Test
Is the impurity mutagenic?ICH M7
classification
Tips for expert assessments
• Look for experimental data
• Vitic is a large, expert-curated database including public sources and member donations
• Read Derek’s detailed commentaries
• These are written by Lhasa’s expert scientists and contain peer-reviewed knowledge
• Use the confidence measures provided both Derek and Sarah
• These apply scientifically robust methodologies which have been published
• They indicate how trustworthy a prediction is and tell you where to focus
• Misclassified and unclassified features…
Features seen in false
negatives
Features not seen in
public datasets
Tips for expert assessments
• Look closely at supporting examples
• Review positive analogues
• could their activity be caused by additional features irrelevant to your compound?
• Review negative analogues
• have they been tested in 5 strains under OECD guidelines?
• Be conservative in your assessments
• If the evidence is not strong enough to convince you..
• treat as positive and consider using purge-based arguments
• run the Ames study
Tips for expert assessments
• You can overturn expert and statistical systems
Worked examples
Compound 1
Compound
Review high level predictions
?
Expert Review
M7 classification
Compound
Review the expert prediction
?
Expert Review
M7 classification
SAR
Alert comments highlight that certain anilines with electron withdrawing groups are excluded from the alert
Mechanism
Alert comments indicate that electron withdrawing groups are expected to reduce the rate of metabolic activation and destabilise intermediates. Therefore, electron withdrawing groups are expected to reduce the mutagenic potential of the aromatic amine.
Review the statistical prediction
?
Expert Review
M7 classification
Almost all close examples are negative
– one positive example is a bis-aniline
Review the statistical prediction
?
Expert Review
M7 classification
The closest analogues in the Sarah Nexus training set contains non-mutagenic
(5 strain) aromatic amines which have relevant substitution patterns compared
to the query compound
Expert review
?
Expert Review
M7 classification
Predicted negative and two relevant nearest neighbours are
non-mutagenic having been tested in five strains
The compound fires an aromatic amine alert.
However, the alert comments indicate that aromatic
amines with electron withdrawing substituents are
known to be non-mutagenic and this relationship is
grounded by mechanistic rationale
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
M7 Classification – Compound 1
?
Expert Review
M7 classification
Class 5
No alerts or alerting with sufficient data to demonstrate lack of mutagenicity
Compound 2
Compound
Review high level predictions
?
Expert Review
M7 classification
Compound
Review the expert prediction
?
Expert Review
M7 classification
Photomutagen mechanism is specific to furocoumarin structure and notrelevant to the query compound. The benzofuran moiety may intercalate;however, benzofuran is non-mutagenic2 and the carboxylic acid is notexpected to react with DNA.
Furocoumarins are known photomutagens which
intercalate with DNA then, upon near-ultraviolet
irradiation, bind pyrimidine bases1.
(1) Venturini et al, Mut. Res., 1981, 88, 17-22; (2) Weill-Thevenet et al, Mut. Res., 1981, 88, 355-362
Review the statistical prediction
?
Expert Review
M7 classification
Activity of analogues is associated with alternative functional
groups: aromatic amine, aromatic nitro.
Review the statistical prediction
?
Expert Review
M7 classification
Most similar compound is a benzofuran with a carboxylic acid at C2
and is negative in multiple strains.
Expert review
?
Expert Review
M7 classification
• Inactive with misclassified features
• ..these are associated with an alternative mechanism that is
specific to furocoumarins and not the query compound
• Positive with 7% confidence
• Activity of nearest neighbours is associated with alternative functional groups
• Most similar example is a non-mutagenic benzofuran
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
M7 Classification – Compound 2
?
Expert Review
M7 classification
Class 5
Compound 3
Compound
Review high level predictions
?
Expert Review
M7 classification
Compound
Review the expert prediction
?
Expert Review
M7 classification
Alert highlights that positive Ames results are observed for acid halides when DMSO isused as the solvent1
Acid halides typically undergo rapid hydrolysis to the carboxylic acid which is a knownAmes negative
(1) Amberg et al, Org. Proc. Res. Dev., 2015, 19, 1495-1506
Review the statistical prediction
?
Expert Review
M7 classification
Activity associated with alternative functional groups: epoxide, N-
nitroso, alkylating agent. Similar acid chloride is negative.
Review the statistical prediction
?
Expert Review
M7 classification
Activity associated with alternative functional groups:
epoxide, N-nitroso, aromatic azide.
Expert review
?
Expert Review
M7 classification
• Positive prediction with equivocal reasoning for acid halide
• Activity of acid halides in the Ames test is known to be a result of using DMSO
• Many acid halides rapidly undergo hydrolysis to the carboxylic acids which are
non-mutagenic
• Negative with 26% confidence
• Activity of nearest neighbours associated with other functional groups
• Most similar acid halide is non-mutagenic
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
M7 Classification – Compound 3
?
Expert Review
M7 classification
Class 5**alternatively use a purge argument to negate the risk of
the acid halide
Compound 4
Compound
Review high level predictions
?
Expert Review
M7 classification
Compound
Review the expert prediction
?
Expert Review
M7 classification
• No alerts identified in DX
• No misclassified or unclassified features
Review the statistical prediction
?
Expert Review
M7 classification
All positive compounds have other features which
may cause the observed activity
Review the statistical prediction
?
Expert Review
M7 classification
• Positive compounds have other features more likely
to cause activity
• Closest analogues fire a DX alert for alpha beta
unsaturated systems. Alkyl substituents at C3 are
excluded from this alert as they have reduced activity due to steric hinderance.
Review the statistical prediction
?
Expert Review
M7 classification
• The closest analogues are negative
• …and have been tested in 5 strains
Expert review
?
Expert Review
M7 classification
• Inactive prediction
• No misclassified features
• Negative prediction
• NN most similar to query compound are negative
• Positive compounds can be explained by presence of other groups
• The alpha beta unsaturated nitrile has shown activity when
unsubstituted (acrylonitrile) but this activates an alert which indicates activation requires a strong EWG and no steric hindrance in order to
react in 1,4 position.
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
M7 Classification – Compound 4
?
Expert Review
M7 classification
Class 5No alerts or alerting with sufficient data to demonstrate lack of mutagenicity
Compound 5
Compound
Review high level predictions
?
Expert Review
M7 classification
10
Review the expert prediction
?
Expert Review
M7 classification
• Inactive prediction
• No misclassified features
• Near-miss of alert 351 for aromatic amine
10
Review the statistical prediction
?
Expert Review
M7 classification
Similar compounds in the training set show inconsistent results (& strain profile)
Nearest examples are positive
10
Expert review
?
Expert Review
M7 classification
• Inactive prediction
• No misclassified features
• Near-miss of alert 351 for aromatic amine
• Positive – 9% confidence
• The nearest neighbour is positive
• Positive compounds do not have any additional toxicophores
10
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
10
M7 Classification – Compound 5
?
Expert Review
M7 classification
Class 3Alerting structure.
Hard to dismiss close potential analogues – suspect we are
near an activity cliff. Recommend testing
10
Compound 6
Review high level predictions
?
Expert Review
M7 classification
Review the expert prediction
?
Expert Review
M7 classification
Alert 307: N-Methylol compound or precursor
N-methylol group matches toxicophore 1.Identified as a DNA reactive centre through hydrolysis to formaldehyde (Ames positive).Covalent DNA binding reported for many N-methylol derivatives.
Review the expert prediction
?
Expert Review
M7 classification
Alert 307: N-Methylol compound or precursor
Toxicophore I positive example
To reflect the frequently low or absent activity of members of this class in
Ames tests the default likelihood has been set as equivocal for non-example
compounds.
Additional information (from Vitic)
?
Expert Review
M7 classification
Substructure search for N-methylol compounds in Vitic 2018.1.0: 20 matches toxicophore I.
Ames overall call positive for 6, negative for 12, conflicting data for 2. Some relevant examples below:
Positive in TA98, TA100 and TA102+/-S9
Positive in TA102+/-S9
Conflicting data in TA98 and TA100+/-S9
*Caution: Ames test may be unresponsive to this class of compounds,
due to low rate of conversion to formaldehyde, itself not highly potent.
Non-standard conditions may be required to observe activity for this class.
Negative examples firing alert 307
Review the statistical prediction
?
Expert Review
M7 classification
Compounds from training set are dissimilar to the test compound.
None contains a N-methylol group matching the toxicophore 1 from the Derek alert 307. No Sarah hypothesis exists for this toxicophore.
Expert review
?
Expert Review
M7 classification
• Active prediction
• Fires alert 307 for N-Methylol compound or precursor
Matches toxicophore 1: N-Methylol identified as DNA
reactive centre through hydrolysis to formaldehyde.
• Inactive prediction
• Compounds from training set dissimilar to the test compound.
None matches the N-Methylol toxicophore
• No Sarah hypothesis exists for this toxicophore.
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
M7 Classification – compound 6
?
Expert Review
M7 classification
Class 3*Alerting structure
*Caution: Ames test may be unresponsive to this class of compounds,
due to low rate of conversion to formaldehyde, itself not highly potent
Non-standard conditions may be required to observe activity for this class.
Compound 7
Review high level predictions
?
Expert Review
M7 classification
Review the expert prediction
?
Expert Review
M7 classification
Alert 027: Alkylating agent
• Chloromethanamine fragment matches alkylating toxicophore (alkyl chloride).
• Alkyl halides identified as electrophilic species capable of directly alkylating DNA.
• This alert describes the genotoxicity of alkylating agents where the carbon
bearing the functional group is a primary or secondary alkyl carbon atom.
Additional information (from Vitic)
?
Expert Review
M7 classification
2 compounds matching the (chloromethyl)dimethylamine pattern with Ames positive calls.
However 2nd toxicophore present for the 2nd structure, which also fires the aromatic nitro alert 329.
Substructure search for the (chloromethyl)dimethylamine pattern
with an Ames overall call in Vitic 2018.1.0
N-(Chloromethyl)-4-nitrophthalimide
Positive in TA98 and TA100 +/-S9
2nd toxicophore present,
fires the aromatic nitro compound alert 329.
Chloromethylphthalimide
Positive in TA98 and TA100 +/-S9
Review the statistical prediction
?
Expert Review
M7 classification
Positive hypothesis for alkyl chloride
Most alkyl chlorides examples are
positive except 3 out of 15
(examples 2, 7 and 9).
Positive hypothesis for alkyl
chloride, toxicophore responsible
for mutagenicity, in agreement with Derek alert 027.
Expert review
?
Expert Review
M7 classification
• Active prediction
• Fires alert 027 for alkylating agents
• Chloromethanamine fragment matches alkyl chloride toxicophore
• Active prediction with 23% confidence
• Matches 2 positive hypotheses
• Alkyl chloride positive hypothesis (most relevant hypothesis),
toxicophore responsible for mutagenicity, in agreement with
Derek alert 027
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
M7 Classification – compound 7
?
Expert Review
M7 classification
Class 3Alerting structure
Compound 8
Expert Review
M7 classification
Review high level predictions
?
Review the expert prediction
?
Expert Review
M7 classification
INACTIVE
• No alerts identified in Derek Nexus.
• No misclassified features identified.
Review the statistical prediction
?
Expert Review
M7 classification
• Derek alert activated requires at least 3 fused
rings for mechanism suggested
• Other positives contained additional toxicophores
Review the statistical prediction
?
Expert Review
M7 classification
most similar compound
• 100% negative in Sarah.
• Inactive in Derek.
• Good strain profile.
Expert review
?
Expert Review
M7 classification
• Inactive prediction
• No misclassified features
• Weak positive prediction (6%)
• Activity of compounds in Sarah are associated with other
functional groups that fire alternative Derek alerts.
• Most similar compound gives negative results
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
M7 Classification – Compound 8
?
Expert Review
M7 classification
Class 5
Compound 9
Review high level predictions
?
Expert Review
M7 classification
Compound
Review the expert prediction
?
Expert Review
M7 classification
Review the expert prediction
?
Expert Review
M7 classification
Review the statistical prediction
?
Expert Review
M7 classification
Expert review
?
Expert Review
M7 classification
• Active prediction with plausible reasoning
• No evidence that could negate prediction
• Ethylene sulfate (similar compound) is active in TA1535
• Out of domain prediction
• Disregarded hypothesis does not have examples that
similar to query compound
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
M7 Classification – Compound 9
?
Expert Review
M7 classification
Class 3 – Alerting structure
Resources to support your expert review
• Selecting your systems
• Distinguishing between expert and statistical systems for application under ICH M7.
Barber, C. Reg. Toxicol. Pharmacol. 2017, 84, 124
• Undertaking expert review
• Establishing best practise in the application of expert review of mutagenicity under ICH M7.
Barber, C.., Reg. Toxicol. Pharmacol. 2015, 73, 367
• Use of in silico systems and expert knowledge for structure-based assessment of potentially
mutagenic impurities. Sutter, A.., Reg. Toxicol. Pharmacol. 2013, 67, 39
• Principles and procedures for implementation of ICH M7 recommended (Q)SAR analyses.
Amberg, A.., Reg. Toxicol. Pharmacol. 2016, 77, 13
• Lhasa’s website has many papers, posters, and videos…
• https://www.lhasalimited.org/library/
Thank you
Head office in Leeds, UKNot-for-profit
Educational charity
A membership organisation
Data & knowledge sharing
Honest broker
Sponsor
PhDs
Teaching
lecturesWork
experience
MIP-DILI
Regulators &
Governmental Agencies(38 are members of Lhasa)
Proprietary data mining
Predictive software
(expert & machine learnt)
Purge
Toxicity Metabolism
Degradation
Undergrad
projects
Publications &
presentations
FDA NIHS eTRANSAFE
eTOXPMDA
MHRAUSP
• Audience voting slide. Use this slide when asking a poll question in GoToWebinar, or a question that
uses the audience voting system.
Lhasa Limited
Granary Wharf House, 2 Canal Wharf
Leeds, LS11 5PS
Registered Charity (290866)
Company Registration Number 01765239
+44(0)113 394 6020
info@lhasalimited.org
www.lhasalimited.org
Work in progress disclaimerThis document is intended to outline our general product direction
and is for information purposes only, and may not be incorporated
into any contract. It is not a commitment to deliver any material,
code, or functionality, and should not be relied upon. The
development, release, and timing of any features or functionality
described for Lhasa Limited’s products remains at the sole
discretion of Lhasa Limited.
Collaborative working disclaimer
The purpose of this meeting is to [INSERT]. In order to avoid a breach of
competition law, participants are reminded not to disclose commercially sensitive or confidential information, or intentions about future market conduct (including
information about products under development, pricing, sales strategy or
customers). This meeting is not intended to lead to any agreement or understanding as to market conduct.
Recommended