mirabilis the scientific future - the... · mirabilis –objective •precompetitive consortium of...

Mirabilis

The Scientific Future

Lhasa Limited ICGM, Basel

13 October 2016Martin Ott

Mirabilis

A tool to estimate the purge (removal) of potentially genotoxic impurities in drugs

• To support the application of ICH M7 guidance

• Option 4

• “Understand process parameters and impact on residual

impurity levels (including fate and purge knowledge) with

sufficient confidence that the level of the impurity in the drug

substance will be below the acceptable limit such that no

analytical testing is recommended for this impurity”

• Develop a knowledge-based approach to provide support,

evidence and guidance for expert assessment

2

Mirabilis – objective

• Precompetitive consortium of 16 companies

• Sharing knowledge and data

• Desire for

• A standardised process with clear definitions of steps

• Application of an accepted scientific approach

• Conservative, robust and trustworthy predictions

• A transparent tool that

• Can be challenged by experts

• Provides evidence that can support good decisions

• Creates reports with direct access to more support if needed

3

Mirabilis – led by a consortium

• Continued scientific guidance and oversight by consortium

• Sharing examples of purge estimates

• Confirm that these are conservative and robust

• Part of the validation of the science

• Will form a publication - 1Q17

• Establishing agreement on

• When an estimated purge is considered sufficient

• Level of information needed to support decision

4

Primary goal for Mirabilis

• Trusted by regulators

• Purge predictions are:

• Conservative

• Consistent/standardised

• Transparent

5

Purge factor calculation

• The threat posed by potential

mutagenic impurities (PMI) arises

from reagents used in synthesis:

• electrophilic reagents (alkylating

agents)

• DNA interchelators

• pro-mutagenic e.g. metabolically

activated

• Any synthetic drug therefore has a

latent MI-related risk

Scheme from Teasdale et al, Org. Process Res. Dev. (2013), 17, 221-230 6

Purge factor calculation

• The original approach: test for the presence of a PMI in

the final API

Very simplistic approach and fails to account for:

• Inherent reactive nature of the PMI

• An understanding the fate of the PMI in the synthetic processes

applied

It is a paradox that the very reactivity that renders the

chemical a concern from a safety perspective can be

the same property that will generally ensure that it is

effectively removed in downstream processes.

7

Scientific prioritisation

1. Reactivity purge predictions

• Development of “knowledge matrix”

• Reaction mining to supplement/support predictions

2. Solubility purge predictions

• Data-driven approach for common compounds

• Solubility prediction (research)

3. Other, e.g. volatility purge predictions8

Scientific prioritisation – Reactivity

• Knowledge gathering to fill matrix

• Prioritised – fill most crucial gaps first

• Coverage (2016 and beyond) agreed with members

• Reaction mining to supplement/support predictions

• In research phase – demonstrate concept first

• Pilot study in preparation to mine member lab data

9

Knowledge matrix (reactivity)

• Matrix

11

Knowledge matrix (reactivity)

• 57 reaction types - 28 prioritised ( 31)

• 15 impurity classes - 4 prioritised

• 124 / 855 (14.5%) “Predicted Reactivity Purge

Factor Knowledge Matrix” points (now called “Cells”)

have been prioritised for 2016

• Speed of development can be expected to increase

in 2017

12

Reactivity prediction & knowledge base

The matrix is implemented as an in silico reaction

knowledge base, which will:

Automatically recognise the structural class of an

impurity

Automatically recognise the transformation in a

synthetic step

Generate a predicted reactivity purge factor based on

empirical data

Present transparent and robust scientific evidence to

support the predicted reactivity purge factor assigned

13

Knowledge base editor – reactivity matrix

Knowledge matrix (reactivity)

14

Knowledge base editor – impurity class

Impurity class

15

Knowledge base editor – transformation class

Transformation class

16

Cell contents

Index

Purge factor

Executive summary

Dependencies

Comments

17

Knowledge base editor – cell contents

Purge factor

Summary

Dependencies

Comments

18

Operation – route editor

Enter route

19

Operation – route editor

Enter route

20

Operation – route editor

Enter route

21

Operation – route editor

Enter route

22

Impurity new impurity

• A reactivity-based purge usually generates a new compound

which may or may not be a impurity of concern

• This does not affect the effectiveness of that purge – the new

compound, if is a pMI, should itself be treated as a new

impurity

• It is the user’s decision whether this impurity should be tracked

• The new impurity should “inherit” any purge value from its

parent

• Mirabilis does not support this yet

• This will be part of a future treatment of concentrations

23

Impurity new impurity (same class)

24

Impurity new impurity (different class)

25

Impurity belonging to multiple classes

• More than one impurity type may be recognised

More than one purge factor prediction – which one to choose?

• It was suggested to choose the more conservative value

“There is still genotoxic/mutagenic potential left”

• However, the compound will be purged by a factor of 100!

• We track the purging of compounds, not of functional groups and/or

genotoxic/mutagenic potential

• Any compounds newly generated by purging reactions should be

in turn considered for tracking, subject to the user’s judgment 26

Conclusion

• The purge tool concept provides a cost-effective, quick and

effective way of assessing the risk posed by a MI in a

regulatory risk assessment.

• The development of an in silico tool provides the basis for

a consistent and systematic cross-industry approach which

is based on expert knowledge.

• Use of these predictions and transparent expert knowledge

aligns directly with principles defined in ICH M7.

27

Conclusion

• Development of the tool is progressing well

• Enthusiastic and committed consortium of 16 members

• Providing scientific guidance and oversight

• Knowledge will be expanded and refined

• Datamining techniques may assist in finding supporting

evidence for reactivity purge factors

• Further research into solubility is planned for the future

28

TimTransformation

Engine

SimonUser

Documentation

MukeshKnowledge

Base

DianeUser Experience

Design

AlanTransformation

Engine

MartinLead

Scientist

NatashaKnowledge

Base MikeKnowledge

BaseSimonProductAnalyst

TomProductTesting

DanKnowledge

Editor

AhmedMirabilisSoftware

PaulLead

Developer

DavidKnowledge

Editor

SuziProject

Management

LizBusiness

Development

Mirabilis 2016 Project Team

SusanneKnowledge

Base

AntonioKnowledge

Base

LaraKnowledge

Base

TonyKnowledge

Base

SamKnowledge

Base

JingMirabilisSoftware

DavidReporting

29

Questions?