Establishing and Maintaining a Product Control System – An Industry Perspective
Mary CromwellDirector
Late Stage Pharmaceutical DevelopmentGenentech
March 23, 2011CaSSS CMC Strategy Forum Europe
Barcelona, Spain
Establishing a Control System for a Monoclonal Antibody Product
Extensive platform knowledge• 5 marketed and legacy products• Clinical development experience with > 40 MAbs
Commonalities• Predominantly IgG type 1• Produced in CHO cells – similar glycan structures• Similar purification processes: 3 column steps• Host cell and downstream process impurities
Differences• Mechanism of action
–Blocking–Effector function requirement–Binding to soluble antigen
• Patient population, route of administration, concurrent therapies
QbD Approach
• Risk-based assessment taking into account– Impact of quality attribute to patient safety and efficacy
> CQA-acceptance criteria (CQA-AC)–Ability of process to control the levels of the attribute
> Predictive ability of small scale models> Observations from at-scale production
–Changes in levels of attribute on stability
• Sensitivity of testing method to attribute, for those that require testing on the control system
Risk Ranking & Filtering Tool for CQA Identification
Does the attribute impact safety or efficacy?Determined by the
available knowledge.
More severe impact higher value.
How confident are we in assigning impact?
Determined by relevance of knowledge.
Higher uncertainty higher value.
Risk that an attribute impacts safety or efficacy
PK
BioactivitySafety
Immunogenicity
Risk = Impact x Uncertainty
Paul Motchnik
Attribute Testing Strategy Risk Ranking and Filtering Tool (ATS RRF)
Quality Attribute
Impact Score
Process or StabilityImpact ScoreX == ATS (1)
(1) Attribute Testing Strategy
Performed for:• DS manufacturing process• DP manufacturing process• DS storage• DP storage
2, 4, 12, 16, 20 1, 2, 4, 10 2-200
Defines Testing Strategy:• No Testing• “Comparability and Monitoring”• Control System (Release, Stability,
and/or In-Process)
Quality Attribute Impact Score
1 Based on a relevant potency assay and dependent on assay variability2 Based on serum exposure (AUC) or FcRn binding. PD considered if information available3 Based on effects observed in clinical studies
Impact & Rating
Biological
Activity1PK2 Immunogenicity3 Safety3
(Potential or Observed)
Very High(20)
>100% change
>40% change on PK
ATA (Anti-Therapeutic Antibodies) detected that may life
threatening
Irreversible or life-threatening AEs and/or life threatening loss of efficacy
High(16)
40-100% change
20-40% change with
impact on PD
ATA detected that may be associated with non-life
threatening loss of efficacy
Reversible AEs and/or loss of efficacy that is not life
threatening
Moderate(12)
20-40% change
20-40% change with no impact on
PD
ATA detected with effect that can be managed by clinical treatment
(i.e. dose titration, medication, etc.)
AE that can be managed by clinical treatment (i.e. dose titration, medication, etc.)
Low(4)
<20% change
<20% change with no impact
on PD
ATA detected with effect on PK or PD, but no effect on safety or
efficacy
Safety or efficacy effect with minimal clinical significance
None (2) No change
No impact on PK or PD
ATA not detected or ATA detected with no effect on PK,
PD, safety, or efficacy
No effect on safety of efficacy
Paul Motchnik
Process Impact Decision TreeQA
Impact Score of 2
or 4?
QAImpact
Score of 2 or 4?
NoStart for each Quality
Attribute
Start for each Quality
Attribute
Yes
AbundanceFilter <1%*
AbundanceFilter <1%*
No
Yes
Sufficiently predictive process model?
Sufficiently predictive process model?
No
Yes
Outcomeof PC/PV &
Linking Studies
Outcomeof PC/PV &
Linking Studies
Result within CQA-TR
Default
Outcomeof PC/PV &
Linking Studies
Outcomeof PC/PV &
Linking Studies
Highly Robust
Process Impact Score = 1
Process Impact Score = 1
Process Impact Score = 10
Process Impact Score = 10
Actual result outside CQA-TR
Process Impact Score = 4
Process Impact Score = 10
Process Impact Score = 10
Process Impact Score = 4
Process Impact Score = 2
Process Impact Score = 2
*Applies to product-related variants only; Abundance threshold for aggregates is < 0.1%; threshold for some sequence variants is <0.2%.
Nathan McKnightAssessment uses data from PC/PV studies, relevant reduced and at scalemanufacturing experience, and worst case linkage studies
Stability Impact Scoring Decision Tree
Start: for each Quality Attribute
Can moleculeform attribute?
Process ImpactScore = 1
Rate of changerelative to CQA-AC
Process ImpactScore = 10
Process ImpactScore = 2
Process ImpactScore = 4
No
Yes
Slow Fast
Moderate
R. Wong
(< 11%*) (>33%*)
(11-33%*)
*of allowable stability range;Assessed to expiry at recommendedstorage temperature and for allowableexcursions
Analysis performed using data collected from qualificationlots and relevant development lots (GMP and non-GMP)
Attribute Testing Strategy RRF
CQA ImpactScore
(CQA RRF)
Process or Stability Impact
ScoreX = Attribute Testing Strategy (ATS) Score
Attribute Testing Strategy (ATS)
Score
< 21 No Testing Required21-50 Comparability & Monitoring (CaM) Testing>50 Control System Testing Required
Note: Highest Impact Score is used for each Critical Quality Attribute (CQA).
CQA Impact Score 1 2 4 102 2 4 8 204 4 8 16 40
12 12 24 48 12016 16 32 64 16020 20 40 80 200
Process/Stab impact
“Comparability and Monitoring” (CaM)
Attribute class to be tested as part of:
• Comparability exercises– Performed to support site transfer, version changes, scale changes– Provides streamlined testing
> Testing includes appropriate (DS or DP) tests designated “CaM” in the Testing Strategy as well as Control System testing (IP, Lot Release, Stability)
> Choice of tests based on risk associated with change; only CaM attributes known to be impacted by particular step that is changing will be tested
• Process Monitoring– Continuous process monitoring
> Subset of CaM attributes > Frequency of monitoring may be attribute dependent
> Control System testing> Key Performance Indicators
ATS Applied to Acidic Variants for MAb OParameter Rationale
QA Impact Score 12 Based on potency impact in CDC and ADCC assays
CQA Acceptance Criteria ≤ 45% Based on level required to maintain potency ≥ 80% at the end of DP shelf life
CQA Target Range for DS Process ≤ 34%
Based on level required to ensure that DP meets acceptance criterion at end of shelf life and a 5% reduction in the CQA-AC
Process/Stability Impact Scores
DS Process 10 Worst-case conditions across all operations result in a batch that exceeds the CQA-TR
DP Process 4 Changes during ambient handling and excursions
DS Stability 2 No changes observed throughout development
DP Stability 10 Significant changes (> 33% of allowable stability range) observed over shelf life and during excursions
Attribute Testing Strategy = QA Impact Score X Process/Stability Impact Score
Attribute Testing Strategy Scores
DS Process 120 Control system testing
DP Process 48 Comparability and Monitoring testing
DS Stability 24 Comparability and Monitoring testing
DP Stability 120 Control system testing
“Acidic variants” is a surrogate CQA for the deamidated, sialylated, and glycatedCQAs that are collectively measured by Ion Exchange Chromatography
Robustness Assessment of Testing Strategy
Testing Strategy Tool used to develop the proposed Control System• Is proposed testing strategy of sufficiently low risk?• Do proposed methods provide adequate control?
Robustness Assessment Tool considers• ATS score (reflecting CQA Impact and Process/Stability control)• Sensitivity of method used for analysis• Testing Strategy (Control System, “CaM”, no testing)
Expected to be iterative process: Unacceptable score indicates that• Testing strategy may need to change• A more sensitive method may be required for testing attribute• Manufacturing process may need to provide greater process control• Shelf-life/allowable excursions may need to be shortened
Robustness Assessment of Testing Strategy
Attribute Testing Score
X TestingStrategy Score
= RobustnessScore
2-200 2, 4, 6, 10 4 - 2000
LR = Lot release; IP = In-Process.
No testing 10
The attribute is monitored directly, collectively, or via a surrogate assay ONLY during “Comparability and Monitoring” testing.
6
The impact of an attribute is measured in a surrogate assay, is measured collectively, or is measured directly with a relatively insensitive assay (IP or LR/Stability).
4
Attribute is measured directly with a suitably sensitive assay (IP or LR/Stability).
2DescriptionScore
Testing Strategy Scoring for Robustness Assessment RRF
LR = Lot release; IP = In-Process.
No testing 10
The attribute is monitored directly, collectively, or via a surrogate assay ONLY during “Comparability and Monitoring” testing.
6
The impact of an attribute is measured in a surrogate assay, is measured collectively, or is measured directly with a relatively insensitive assay (IP or LR/Stability).
4
Attribute is measured directly with a suitably sensitive assay (IP or LR/Stability).
2DescriptionScore
Testing Strategy Scoring for Robustness Assessment RRF
Score </= 400 indicates Robust Control Strategy
Attribute Testing Strategy Robustness (ATSR) Scoring Matrix
ATS Score
Testing Score
2 4 8 12 16 20 24 32 40 48 64 80 120 160 200
2 4 8 16 24 32 40 48 64 80 96 128 160 240 320 400
4 8 16 32 48 64 80 96 128
160 192 256 320 480 640 800
6 12 24 48 72 96 120 144 192
240 288 384 480 720 960 120
0
10 20 40 80 120 160 200 240 320
400 480 640 800 120
01600
2000
ATSR Scoring
< 400 Robust Testing Strategy> 400 Non-Robust Testing Strategy
Robustness Assessment for Afucosylation for MAb O: Glycan Assay Compared to ADCC Potency Assay
• Control of afucosylation by CE-glycan assay is Low Risk
Attribute Step ATS
Score Testing Strategy
Test Method
Testing Strategy
Score Robustness
Score Risk Level
DS Process 200 Control
System 2 400
DP Process 20 No testing 10 200
DS Stability 20 No testing 10 200
G0−F
DP Stability 20 No testing
CE Glycan Assay
10 200
Low
ATS = Attribute Testing Strategy; CE = capillary electrophoresis; DP = Drug Product; DS = Drug Substance.
Robustness Assessment for Afucosylation for MAb O: Glycan Assay Compared to ADCC Potency Assay
• Control of afucosylation by ADCC potency assay ishigher risk
• high impact CQA• low process control
Attribute Step ATS
Score Testing Strategy
Test Method
Testing Strategy
Score Robustness
Score Risk Level
G0−F DS Process 200 Control
System
ADCC Potency Assay
4 800 High
ATS = Attribute Testing Strategy; CE = capillary electrophoresis; DP = Drug Product; DS = Drug Substance.
Minimum Control System
Implemented to • Safeguard against the unpredicted• Provide some measure of consistency
Drug Substance Control System• COC• CE-SDS as a General Impurity assay• Charged variant assay (IEC/icIEF) or Hydrophobic variant assay (eg HIC,
RP-HPLC)
Drug Product Control System• SEC
In Process vs Lot Release (End Product)• if measures CQA for particular product, then on Lot Release
– Acceptance criterion
• if does not monitor CQA, then placed as In Process test– Acceptance criterion based on worst case linkage study results– Action limit
Tests may be added to the stability program
Summary of Testing Strategy for Drug Substance for MAb O and MAb P
Control System Testing
Attribute In Process Lot Release Stability
Process Monitoring and Comparability No Testing
Product-Related Variants Afucosylated glycans O P
Glycan distribution O P
Fragments Ob,Pb
Aggregates Ob,Pb
Acidic variantsc Pa O Ob
Non-glycosylated heavy chain
O P
CDR Oxidation O Ob P
M430 Oxidation Ob,Pb
Other Oxidation O, P
Reduced MAb O, P
Protein Conformation O, P
Cys22-Cys96 free thiol O, P
N-terminal extension O, P
C-terminal variants O, P
Appearance Oa,Pa aDenotes information from Minimum Control System assay. b Denotes attribute should be tested on stability for Comparability. c Acidic variants” includes CDR deamidation, Fc deamidation, glycation, and sialylation.
Summary of Testing Strategy for Drug Substance for MAb O and MAb P
Control System Testing
Attribute In Process Lot Release Stability
Process Monitoring and Comparability No Testing
Statutory Requirements/Stability-Impacting Attributes (cont'd) Identity O, P
Protein Concentration O, P
Osmolality O, P
pH O, P
Acetate Concentration O
Trehalose Concentration O
Histidine Concentration P
Sucrose Concentration P
Polysorbate Concentration
O, P
a Denotes information from Minimum Control System assay.
Summary of Testing Strategy for Drug Substance for MAb O and MAb P
Control System Testing
Attribute In Process Lot Release Stability
Process Monitoring and Comparability No Testing
Process-Related Impurities Mycoplasma O, P
Rodent Parvovirus O, P
CHOP O, P
General Impurity Oa,Pa
Leached Protein A O, P
DNA O, P
Bioburden O, P
Endotoxin O, P O, P
Raw Materials O, P
a Denotes information from Minimum Control System assay.
Summary of Testing Strategy for Drug Product for MAb O and MAb P
Control System Testing
Attribute In Process Lot Release Stabil ity
Process Monitoring and Comparability No Testing
Product-Related Variants Afucosylated glycans O, P
Glycan distribution O, P
Fragments Oa, P P Ob
Aggregates Oa, Pa P Ob
Acidic variantsc O, Pa O, Pa
Non-glycosylated heavy chain
O, P
CDR Oxidation Ob P
M430 Oxidation Ob, Pb
Other Oxidation O, P
Reduced MAb O, P
Protein Conformation O, P
Cys22-Cys96 free thiol O, P
N-terminal extension O, P
C-terminal variants O, P a D e n o te s inf orm a tio n fr o m M in im u m C o ntro l S y s te m a s s a y.
b D e n o te s attr ib u te s ho u ld b e te s te d o n s ta b ility fo r C o m pa r a b ility . c A c id ic v ar ia n ts ” in c lu de s C D R de a m id a t io n, F c d e am id a t io n , gl yc atio n , a n d s ia l y la ti on .
Summary of Testing Strategy for Drug Product for MAb O and MAb P
Control System Testing
Attribute In Process Lot Release Stability
Process Monitoring and Comparability No Testing
Process-Related Impurities Mycoplasma O, P
Rodent Parvovirus O, P
CHOP O, P
Leached Protein A O, P
DNA O, P
Bioburden O, P
Endotoxin O, P O, P
Raw Materials O, P
Leachables O, P
Summary of Testing Strategy for Drug Product for MAb O and MAb P
Control System Testing
Attribute In Process Lot Release Stability
Process Monitoring and Comparability No Testing
Statutory Requirements/Stability-Impacting Attributes Identity O, P
Protein Concentration O, P
Osmolality O, P
pH O, P
Acetate Concentration O
Trehalose Concentration O
Histidine Concentration P
Sucrose Concentration P
Polysorbate Concentration
O, P O, P
Appearance O, P O, P
Visible Particles O, P O, P
Subvisible Particulates O, P O, P
Fill Volume O, P
Container closure O, P
Sterility O, P
Lifecycle Management of the Control System
• Testing strategy needs to be reassessed based on• Revision of CQA Impact score and/or acceptance criteria changes based on:
– Availability of additional information (clinical, nonclinical, characterization)– Will not be performed as a function of manufacturing history
• Revision of Process Impact Scores based on:– Response to findings from continuous process verification or control
space management– Additional experimental work to expand design space (requires health
authority approval)• Continuous method innovation
• Commitment to assess testing strategy on a periodic basis• Verify that Design Space CQA-TR have been appropriately established• Utilize information from continuous process monitoring as well as product
trending• Determine if Control System needs to be modified
• Resulting changes to the Control System will require health authority approval
Summary
Risk-based tools developed to define Testing Strategy and Control System• Utilizes knowledge of
– CQA Impact on potency, immunogenicity, safety, and PK/PD– Process Impact – Stability Impact– Specificity and Sensitivity of analytical method
• Assigns attributes to three categories of testing– Control System (Lot release, stability, in-process)– “Comparability and Monitoring”– No testing
• Robustness assessment for control of CQAs– Iterative process– Evaluates risk that control strategy is insufficient
Summary, continued…
Use of filters in assessment of testing strategy• Abundance filter as part of Process Impact, not CQA ID• Default process impact score for low impact quality attributes
– Balances wide CQA Acceptance Criteria with potentially little process impact knowledge
Application of Minimum Control Strategy• Safeguard against the unexpected
Lifecycle Management of Control System• Performed periodically• May be revised based on additional information regarding
– CQA Impact– Process/Stability Capability– Method improvements
Advantages of the QbD Approach to Developing a Control Strategy
Clear, logical approach – applicable to all types of biologics
Control Strategy development relies on significant process and product knowledge
• Clear rationale for selection of attributes to test and the type of testing strategy applied can lead to streamlined Control System testing
• Understanding impact of each process step to CQA levels leads totargeted testing for comparability
– Only test those attributes impacted by changed steps
Acknowledgements
Jerry DongLynn Gennaro Yung-Hsiang KaoParamjit KaurDaniel KelatiBrian KelleyLynne KrummenReed HarrisKathy HsiaRaquel IversonKim Latimer
Nadja Alt (Roche)Bernd Hilger (Roche)
Joseph MarhoulNathan McKnightPaul MotchnikDave ReifsnyderSofia RibeiroNatalie Saldou-HoltzCristina SanchezDieter SchmalzingRon TaticekPin-Yee WongRita Wong
A-MAb Case Study: Elements of the Control Strategy
Raw Material Control
Procedural Controls
Process Parameter Controls
In-Process Testing
Specifications (Lot Release and Stability)
Characterizationand Comparability Testing
Process Monitoring
Process Control
Testing
Continual Process Verification
Today’s Discussion
CDR Deamidation of MAb P
CDR Deamidation (N52, N54, N61, N99) Parameter Rationale QA Impact Score 16 Potential impact to potency CQA Acceptance Criterion Not Determined CQA-Target Range for DS Process
Not Determined
Allowable Stability Range Not Determined
Unable to generate species to assess in potency assay
DS Process 1 DP Process 1 DS Stability 1 Process/Stability Impact Scores
DP Stability 1
Abundance is <1% in Phase III material; could not generate through forced degradation at high pH
Attribute Testing Strategy = QA Impact Score X Process/Stability Impact Score DS Process 16 DP Process 16 DS Stability 16 Attribute Testing Strategy Score
DP Stability 16
No Testing
Met 430 Oxidation in MAb P
Fc Oxidation (M430) Parameter Rationale QA Impact Score 12 Potential impact to PK CQA Acceptance Criterion ≤ 10% Based on distribution of
acceptance criteria for attributes that may impact PK
CQA-Target Range for DS Process
≤ 9.5% Standard 5% reduction from CQA-AC
Allowable Stability Range 9% Difference from the level generated in the DS process and the CQA-AC
DS Process 2 Worst case results well within CQA-TR
DP Process 2 Worst case results well within CQA-TR
DS Stability 2 Not observed during frozen storage
Process/Stability Impact Scores
DP Stability 4 Increases 0.04%/month; ≤ 2% expected over shelf-life
Attribute Testing Strategy = QA Impact Score X Process/Stability Impact Score DS Process 24 DP Process 24 DS Stability 24 Attribute Testing Strategy Score
DP Stability 48
Comparability and Monitoring Testing
Leached Protein A in MAb P Leach Protein A
Parameter Rationale QA Impact Score 12 Potential impact to safety and
immunogenicity CQA Acceptance Criterion ≤ 40 ppm Based on Literature information
with ½ log reduction for safety factor
CQA-Target Range for DS Process
≤ 38 ppm Standard 5% reduction from CQA-AC
Allowable Stability Range Not applicable NA DS Process 2 Worst case linkage study results
well within CQA-TR DP Process 1 DS Stability 1
Process/Stability Impact Scores
DP Stability 1
Not impacted by DP manufacturing process or storage
Attribute Testing Strategy = QA Impact Score X Process/Stability Impact Score DS Process 24 Comparability and Monitoring
Testing DP Process 12 DS Stability 12
Attribute Testing Strategy Score
DP Stability 12 No Testing
Lifecycle Management of the Control System
No Testing Control SystemComparability &Monitoring
Updated ATS RRFAssessment
Updated Stability ImpactScores
Updated Process Impact Scores
Updated CQAImpact Score
New Clinical and Nonclinical Information
QA Inputs toOriginal PC/PV RRFAssessment for Step
Being Changed
Comparability Attributes
REV
ISED
New Product-SpecificKnowledge
New PlatformKnowledge
Phase I Clinical Development: Designation of Presumptive CQAs (pCQAs)
Research or platform knowledge of impact to safety, immunogenicity, potency, or pharmacokinetics
Sequence information –• hot spots for oxidation or deamidation in the CDR or FcRn binding region?
• Information from mutant studies performed during molecule selection
Target product profile: patient population, route of administration
Phase I Clinical Development: Platform Specifications
Minimum set of tests and acceptance criteria for DS and DP• Platform methods used, if possible
Acceptance criteria• Statutory requirements
–Eg subvisible particulates must comply with EP, USP• Platform acceptance criteria for attributes that may impact safety
–Eg Minimum % monomer allowed; indirectly controls for higher molecular weight species
• Relative acceptance criteria for attributes that may impact potency
–Based on the level of the attribute present in the material used in the toxicology study to support entry in humans
–Sufficiently wide to account for limited manufacturing capability, limited assay variability knowledge
Phase III Clinical Development
• List of quality attributes is re-examined to define CQA categorization based on
– extensive analytical characterization– in vitro assessments of biological activities, binding to FcRn– results from clinical studies to date
• Molecule specific analytical methods are developed and validated to test particular attributes
• Specifications are defined based on understanding of– impact of quality attribute on patient safety and efficacy– degradation observed during storage of DS and DP– ability of process to control variants