aFucosylation

Report
Control Strategy for Glycosylation
Using a QbD Approach:
Monoclonal Antibody with Effector
Function from the A-Mab Case Study
CMC Forum Washington, DC
Workshop I - CQAs
July , 2010
Presented by Victor Vinci, Eli Lilly
CMC BWG – A-Mab Case Study
Working Group Members
•
•
•
Amgen Team: Joseph Phillips (Lead), Bob Kuhn
Abbott Team: Ed Lundell (Lead), Hans-Juergen Krause, Christine Rinn, Michael
Siedler, and Carsten Weber
Eli Lilly Team: Victor Vinci (Lead), Michael DeFelippis, John R Dobbins, Matthew
Hilton, Bruce Meiklejohn, and Guillermo Miroquesada
Genentech Team: Lynne Krummen (Lead), Sherry Martin-Moe, and Ron Taticek
GSK Team: Ilse Blumentals (Lead), John Erickson, Alan Gardner, Dave Paolella,
Prem Patel, Joseph Rinella, Mary Stawicki, Greg Stockdale
MedImmune Team: Mark Schenerman (Lead), Sanjeev Ahuja, Laurie Kelliher ,
Cindy Oliver , Kripa Ram, Orit Scharf, and Gail Wasserman
Pfizer Team: Leslie Bloom (Lead) and Amit Banerjee, Carol Kirchhoff, Wendy
Lambert, Satish Singh
Facilitator Team: John Berridge, Ken Seamon, and Sam Venugopal
•
Plus help from many others
•
•
•
•
•
Vinci/Defelippis - CMC BWG
QbD Case Study
Lilly - Company Confidential 2010
2
Creating a Biotech Case Study:
“A-Mab”
• Based on a monoclonal
antibody drug substance and
drug product
–
–
–
–
–
–
“A-Mab”
Humanized IgG1 (w/ effector function)
IV Administered Drug (liquid)
Expressed in CHO Cells
Treatment of NHL
Molecule designed to maximize
clinical outcomes and minimize
impact on quality attributes (TPP)
• Publically and freely available
as a teaching tool for industry
and agencies at CASSS or ISPE
Vinci/Defelippis - CMC BWG
QbD Case Study

Why Monoclonal Antibody?
 Represents a significant number
of products in development
 Good product and process exp.
in dev. & manufacture
 Reasonable level of complexity
Lilly - Company Confidential 2010
3
QbD Development Paradigm
Creation of a Control Strategy
Animal In-Vitro
Studies Studies
Input Material Controls
High Criticality
Attributes
Product Quality
Attributes
Procedural Controls
1.Quality attributes to be
considered and/or controlled
by manufacturing process
Criticality
Assessment
2. Acceptable ranges for
quality attributes to ensure
drug safety and efficacy
Process Targets
for Quality
Attributes
Process
Development and
Characterization
Design
Space
Control Strategy Elements
Safety and
Efficacy Data
Process Controls
Continuous Process Verification
Prior
Clinical
Knowledge Studies
Process Parameter
Controls
Testing
In-Process Testing
Specifications
Characterization &
Comparability Testing
Attributes that do not need to
be considered or controlled
by manufacturing process
Process Monitoring
Low Criticality
Attributes
Product Understanding
Vinci/Defelippis - CMC BWG QbD Case
Study
Process Understanding
Lilly - Company Confidential 2010
4
CQA Risk Ranking & Filtering Tool
A Continuum of Criticality (Tool #1 Ex.)
• Assess relative safety and efficacy risks using two factors:
– Impact and Uncertainty used to rank risks
• Impact = impact on safety or efficacy, i.e. consequences
– Determined by available knowledge for attribute in question (prior, clinical, etc)
– More severe impact = higher score
• Impact on biological activity, PK/PD, immunogenicity, adverse effects
• Uncertainty = uncertainty that attribute has expected impact
– Determined by relevance of knowledge for each attribute
– High uncertainty = high score (no information with variant or published lit. only)
– Low uncertainty = low score (data from material used in clinical trials)
Severity = Impact x Uncertainty
• Severity = risk that attribute impacts safety or efficacy
Vinci/Defelippis - CMC BWG
QbD Case Study
Lilly - Company Confidential 2010
5
Criticality Ratings for Glycosylation
Attribute
Criticality
Aggregation
60
aFucosylation
60
Galactosylation
48
Deamidation
4
Oxidation
12
HCP
36
DNA
6
Protein A
16
C-terminal lysine
variants (charge
variants)
4
Glycoslyation - High Criticality
• Example is for afucosylation
and galactosylation; other
glycan structures require
individual consideration
• Primarily impacted by
production BioRx
• No clearance or modification in
DS
• Not impacted by DP process or
stability
Note: Assessment at beginning of development
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
6
Platform and Product Specific Experience
Attribute
Galactose
Content
aFucosylation
Prior
Knowledge
Clinical
experience of 1040% G0 for YMab, another
antibody with
CDC activity as
part of MOA; no
negative impact
on clinical
outcome
In-vitro
Studies
0-100% gal
content has
statistical
correlation w/
CDC activity w/
A-Mab
Studies show that
100% G0 or
100%G2 have
comparable
ADCC
1-11%; Clinical
A-Mab with 2experience with
13% afucosylation
X-Mab and Ytested in ADCC
Mab; both X-Mab assay; linear
and Y-Mab have correlation; 70ADCC as part of 130%
MOA
Vinci/Defelippis - CMC BWG QbD Case
Study
Non-clinical
Studies
Clinical
Experience
Claimed
Acceptable
Range
No animal studies 10-30%
10-40%
Animal model
available;
modeled material
(15%) shows no
significant
difference from
5%
2-13%
Lilly - Company Confidential 2010
5-10%;
Phase II and
Phase III
7
CQA Linkage to Process Knowledge
afucosylation and galactosylation are assigned as CQAs due to linkage to
ADCC and CDC activity and proposed NHL therapeutic need
Analytical characterization method for afucosylation and galactosylation is
CE-LIF:
Bioassay development led to a robust assay with a linear correlation between
aFuc (2-13%) and ADCC activity (bioassay range of 70 – 130%)
Bioassay for CDC showed no impact over the range of galactosylation (10 – 40%)
produced in clinical material
Ranges of afucosylation and galactosylation can be ensured by control of
bioreactor process parameters found to have influence on these structures.
Release testing with Biopotency assay for drug product (acceptance
criterion 70 – 130%) confirms appropriate product quality
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
8
Influence of Glycosylation on ADCC and
CDC Effector Functions
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
9
Experimental Design
Progression of Studies for Production Bioreactor
Prior knowledge and risk assessments inform designed experiments:
• Risk analysis tools guide informed assessments
• Risk assessment links product attributes with parameters
• DOE’s allow understanding of the impact of process parameters and
attributes
• Risk assessments are iterative and continue through the lifecycle of
product
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
10
Risk Assessment Approach
Multiple Assessments Throughout the
A-Mab Development Lifecycle for Entire Process
Process 2
Quality
Attributes
Process 1 2
Life Cycle
Management
Design Space
Prior Knowledge
Process Understanding
Process
Development
Process
Characterization
Product Understanding
Draft Control
Strategy
Process
Performance
Verification
Final Control
Strategy
Process
Parameters
Risk
Assessment
Risk
Assessment
Risk
Assessment
Risk
Assessment
You Are Here
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
11
Example of Risk Assessment Tool
Approach to Process Characterization
Step 1. Use a Fish-bone (Ishikawa) diagram to identify parameters and attributes that
might affect product quality and process performance
Agitation
Production
Bioreactor
N-1 Bioreactor
In Vitro Cell
Age
Seed
Seed Density
Viability
Temperature
Shear/
Mixing
Working
Volume
Harvest
# of
Impellers
CO2
DO
Control
Parameters
Scale
Effects
pH
Nominal
Vessel Volumne
Design Impeller
Design
Duration
Baffles
Procedures
Gas
Transfer
Temperature
pH
Airflow
Sparger Design
Aggredates
Fucosylation
Galactosylation
CEX AV
HCP
DNA
Antifoam
Time of Feeding
Filtration
Volume of
Feed
Operations
Amount Delivered
Storage
Temperature
Concentration
Preparation
pH
Pre-filtration
hold time
Procedures
Age
Number of
Feeds
Age
Operations
Procedures
Storage
Temperature
Pre-filtration hold
time
[Antifoam]
Age
[NaHCO3]
Timing
Preparation
Osmolality
Filtration
Feed
Vinci/Defelippis - CMC BWG QbD Case
Study
[Glucose]
Glucose Feed
Lilly - Company Confidential 2010
Medium
Concentration
12
A-Mab: Mid-Development Risk Assessment Approach
Rank parameters and attributes from Step 1 based on severity of impact and control capability.
Identify interactions to include in DOE studies
R isk M itig atio n
T urbidity at
harvest
V iability at
H arvest
P roduct Y ield
DNA
P ro cess A ttrib u tes
HCP
D eam idation
G alactosylation
aF ucosylation
P ro cess P aram eter in
P ro d u ctio n B io reacto r
A ggregate
Q u ality A ttrib u tes
Inoculum V iable C ell C o nc entr
Inoculum V iability
DOE
Linkage S tudies
Inoculum In V itro C ell A ge
E O P C S tudy
N -1 B io reactor pH
Linkage S tudies
N -1 B io reactor T em pe rature
Linkage S tudies
O sm olality
DOE
A ntifoam C o ncentration
N ot R e quired
N utrient C oncen tration in
m edium
DOE
M edium storage te m pe ratu re
M edium H old S tudies
M edium hold tim e befo re
filtration
M edium H old S tudies
M edium F iltration
M edium H old S tudies
M edium A ge
M edium H old S tudies
T im ing of F eed addition
N ot R e quired
V olum e of F eed addition
DOE
C om pon ent C oncentration in
F eed
DOE
T im ing of glucose feed
addition
D O E -Indirect
A m ount of G lucose fed
D O E -Indirect
D issolved O xygen
DOE
D issolved C arbon D ioxide
DOE
T em peratu re
DOE
pH
DOE
C ulture D uration (days)
DOE
R em nan t G lucose
C oncentration
Vinci/Defelippis - CMC BWG
QbD Case Study
Potential impact to
significantly affect a
process attribute
such as yield or
viability
D O E -Indirect
Lilly - Company Confidential 2010
Potential impact to QA
with effective control of
parameter or less
robust control
Note: pH is red or
critical at this stage due
to linkage to
glycosylation
13
MCC Bioreactor
Control Strategy Elements by System - pH
Raw Materials (Reg/QMS) – vendor qualification; media (or buffer) make-up based on
instructions, weight based; pH check post make-up
Equipment (QMS) – bioreactor design (probe type/placement), probe vendor
qualification, receipt verification, linked to IQ/OQ and PV for bioreactor
Automation (QMS) – control loop qualified (CSV) and controlled via DCS, alert/action
alarms aligned with process, data monitored continuously and archived
DOE and Models (QMS/Reg) – small-scale models use parameter ranges intended for
large-scale; confirm during pivotal and commercial tech transfer
In Process/Operations (QMS) – pH probe calibration (pre-run), batch record
instructions on how to do daily check and adjustment, data trended
Specification Limits/Tests (Reg/QMS) – Control Strategy in place, validated methods
reflecting QbD analytical development
Process Verification/Continuous Monitoring (QMS/Reg) – MVA (PLS) or SPC
monitoring of performance over manufacturing lifecycle
Vinci/Defelippis - CMC BWG
QbD Case Study
Lilly - Company Confidential 2010
14
Continuity of Ranges
Attributes and Parameters in Study
Levels of CQAs:
CQA
Afucosylation (%)
Galactosylation (%)
Lower Limit
2
10
Higher Limit
13
40
Parameter Ranges:
Platform (2 liters and at-scale FHD)
Screening Study (Central Composite)
Design Space Proposal
Batch Record (Pivotal and Comm.)
Automation Alarms
pH 6.6 – 7.1 (initial set pt*)
pH 6.6 – 7.1 (2 liter)
pH 6.6 -7.1 (commercial)
pH 6.95
(initial ref pt)
pH 6.85 lo/pH 6.95 hi alert-control space
pH 6.7 lo/pH 7.1 action-design space
*Note that pH variable is set at initial as ref pt and moves through low (base) and high (acid or
CO2) control
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
15
DOE Studies to Define Design Space
Bringing Together Process and Product Attributes
Example of DOE Results from Screening Study (Process 2). N=20.
Prediction Profiler
Titer (g/L)
3.743131
±0.076052
5
4
HCP (ppm) Galactosylation aFucosylation
695538
(%)
6.439933
±16518.3
29.28939
±0.226948
±0.674582
3
8
6
4
32
28
24
2500
2000
1500
32
28
24
3.0
2.6
2.2
35
Temperature
(C)
50
DO (%)
Vinci/Defelippis - CMC BWG QbD
Case Study
100
CO2 (%)
6.85
pH
1.2
[Medium]
(X)
400
Osmo (mOsm)
Lilly - Company Confidential 2010
12
Feed (X)
1
IVCC (e6
cells/mL)
19
-0.1
.1
.3
.5
.7
18
17
16
440
9
10
11
12
13
14
15
.7
.8
.9
1
1.1
1.2
1.3
15
420
400
380
1.6
360
1.4
1.2
1
7
7.1
.8
6.9
6.8
6.7
70
40
60
80
100
120
140
160
6.6
60
50
40
36
30
35
35.5
34.5
1.8
34
Aggregates
(%)
2.515119
±0.03524
CEX % Acidic
Variants
27.66898
±0.480814
DNA (ppm)
1935.343
±89.55908
1e+6
8e+5
6e+5
4e+5
17
Duration
(d)
0.21
Curvature
16
Moving Toward Design Space
Follow-up Studies and Analysis
Augment the screening design to enable estimation of a full response surface:
all main effects
two-way interactions
quadratic effects
Additional runs form Central Composite Design (when comb. w/ previous runs):
8 additional runs form full factorial on important parameters.
8 axial points allow to estimate non-linear relationships
4 parameters and 6 QA’s (responses)
N=40 total bioreactor runs
(4 blocks of 10, ~12 weeks)
8 center points total
Response surface model captures all input – output
relationships and is suitable to define the design space
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
17
Contour Profi le r
Horiz
V ert
Contour Profi le r
Factor
Curr ent X
Temperature (C)
Horiz
V ert
35
DO (% )
Factor
Curr ent X
Temperature (C)
50
CO2 ( %)
34.999293
DO (% )
100
50
Develop Multivariate Models to define Design Space
pH
CO2 ( %)
6.85
[Medium] (X)
1.2
Osmo lality (mOs m)
IVCC (e6 cells/mL)
aFucosylation
Galac tosylation (%)
40
HCP ( ppm)
15000000
DNA (ppm)
19000
40
A ggregates ( %)
3
7.1
Hi Limit
15
Response
.
Conto ur
Curr entTiter
X ( g/L)
11
40
20
HCP
70 ( ppm)
.
15000000
DNA (ppm)
6.85
40
12
IVCC (e6 cells/mL)
Cultur e Duration ( days)
Response
Temperature (C)
DO (% )
6.8
CO2 ( %)
pH
6.7
>11%
[Medium] (X)
Osmo lality (mOs m)
6.6
aF ucosylati on
>11%
Feed (X)
34e Duration
34.5
Cultur
( days)
35
35.5
36
Temperature (C)
40
15000000
40
3
CEX % A cidic V ariants
26.7
20
40
3
1.3
.
3
2
20
1465.4
.
.
20
40
.
Curr ent X
35
40
50
15000000
40
.
6.85
40
1.2
3
440
33.1
3
1.3
1
35.5
15
36
Conto ur
Temperature (C)
aFucosylation
40
Galac tosylation (%)
CEX % A cidic V ariants
1479.0
35
DO (% )
50
pH
CO2 ( %)
34.5
35
35.5
Horiz
36
Temperature (C)
20
40
1.3
.
3
6.8
Gal ac tosylati on (%)
Feed (X)
6.7
12
IVCC (e6 cells/mL)
6.7
1
>40%
Cultur e Duration ( days)
6.6
Titer ( g/L)
34
aFucosylation
34.5
35
35.5
36
Temperature (C)
Galac tosylation (%)
Curr ent Y
0
Horiz
V ert
2
Lo Limit
.
29.1
DO
(% )
15000000
Hi Limit
5.1
Factor
6.3
Temperature
(C)
20
HCP ( ppm)
2
11
40
702394.3
CO2 ( %)
19000
CEX % A cidic V ariants
.
1965.8
pH
40
A ggregates ( %)
.
.
20
40
1.6o lality (mOs m) .
Osm
3
28.4
[M
edium] (X)
3
15000000
Feed (X)
7.1
Cultur e Duration ( days)
pH
Temperature (C)
DO (% )
6.8
CO2 ( %)
pH
6.7
[Medium] (X)
Osmo lality (mOs m)
Feed (X)
6.6
Cultur e Duration ( days)
34
34.5
35
Response
35.5
Conto ur
2
mmHg
100 mmHg
HCP ( ppm)
CO2
DNA (ppm)
40
15000000
40 mmHg
CEX % A cidic V ariants
19000
40
3
Hi Limit
.
11
33.5
6.8
669715.6
20
1973.9
.
.
32.9
20
40
1.5
.
3
3
40
.
Factor
7 Temperature (C)
34.5
.
.
40
26.9
20
40
3
1.6
.
3
Curr ent X
35
40
50
40
.
6.85
40
1.2
3
440
12
1
35
35.5
17
36
Temperature (C)
Conto ur
Curr ent Y
2
Lo Limit
4.9
20
15000000
35
35.5
Horiz
36
V ert
7
2
20
1961.3
.
40
30.7
20
3
1.6
.
34.5
35.5
Temperature (C)
Galac tosylation (%)
pH
2
15000000
DNA (ppm)
.
4.3
V ert
.
2459.8
DO (% )
40
A ggregates ( %)
28.6( %)
CO2
3
.
.
20
40
.
3
1.8
pH
[Medium] (X)
7.1
Feed (X)
CO2 ( %)
pH
Osmo lality (mOs m)
Feed (X)
Cultur e Duration ( days)
34
34.5
Gal ac tosylati on (%)
35
35.5
Curr ent Y
aFucosylation
mmHg
100 mmHg
CO2
DNA (ppm)
5.3
2
11
20
30.2
6.9
870128.1
20
2482.5
.
.
40 mmHg
CEX % A cidic V ariants
32.8
20
40
1.8
.
3
40
6.8
3
.
40
35
50
40
1.2
30.1
Osm
o lality (mOs m) 20
40
440
(%)
1
Titer ( g/L)
0
aFucosylation
2
20
15000000
CEX % A cidic V ariants
A ggregates ( %)
7
35.5
6.9
pH
pH
<20%
36
6.8
6.7
6.7
6.6
5.4
All other CQAs did not exceed Quality Limits
when process operated within Knowledge
Space & Design Space
.
Lo Limit
3
Hi Limit
.
5.2
2
27.3
20
.
11
40
.
15000000
2443.6
.
.
40
30.5
20
40
3
1.9
.
3
*Note that DO and Feed Conc from earlier study
are controlled in same range
7.1
35
Temperature (C)
Curr ent Y
15000000
889758.9
19000
6.6
34.5
19
Conto ur
DNA (ppm)
7
12
36
Gal ac tosylati on (%)
6.8
40
6.85
HCP ( ppm)
6.7
34
20
1.9
40
Galac tosylation (%)
15000000
7.1
6.9
.
27.0
11
3Gal ac tosylati on
40
3
Curr ent X
35.5
.
11
40
15000000
.
Hi Limit
.
2
20
.
.
1.8
Feed
(X)
Lo Limit
4.7
21.1
2434.0
19000
Response
2
19000
A ggregates ( %)
6.6
3
.
.
IVCC (e6 cells/mL)
34
34.5
35
Cultur e Duration ( days)
Temperature (C)
Hi Limit
.
15000000
.
.
40
898827.7
.
2
15000000
4.9
20
Hi Limit
20
<20%
3
Lo Limit
5.1)
DO (%
25.9
CO2
( %)
40
.
7
Curr ent Y
2
15000000
.
900138.3
pH
6.7
2528.5
[Medium] (X)
19000
6.1
pH
0
HCP ( ppm)
6.8
20
15000000
1
Conto ur
Titer ( g/L)
160
Galac tosylation (%)
2
Lo Limit
5.4
Temperature
(C)
19
7.1
36
Temperature (C)
6.9
CurrFactor
ent Y
V ert
Conto
ur
Horiz
0
360
CEX % A cidic V ariants
A12
ggregates ( %)
IVCC (e6 cells/mL)
Response
440mOsm
40 tosylation (%)
Galac
6.85
HCP ( ppm)
1.2 (ppm)
DNA
<20%
[Medium] (X)
36
71
400mOsm
35 ( g/L)
Titer
50
aFucosylation
pH
360mOsm
DO (% )
35.5
19 Profi le r
Contour
Cultur e Duration ( days)
Curr entResponse
X
Temperature (C)
35
12
Osmolality
IVCC (e6 cells/mL)
< 2%
Factor
1
19
0
34.5
1.2
7.1
400
Osmo lality (mOs m)
7
12
Conto ur
Galac tosylation (%)
Temperature (C)
Curr ent X
HCP ( ppm)
35
DNA (ppm)
50
CEX % A cidic V ariants
70
A ggregates ( %)
6.85
40
15000000
aF ucosylati on
Contour Profi le r
440
IVCC (e6 cells/mL)
34
aFucosylation
11
20
Design Space for Culture Duration 19 Days
CEX % A cidic V ariants
Osmo lality (mOs m)
Titer ( g/L)
.
2
24.8
Factor
891294.0
Temperature (C)
19000
1.2
6.6
Response
Hi Limit
5.4
20
Horiz
HCP ( ppm)
50
100
6.85
[Medium] (X)
Cultur e Duration ( days)
Lo Limit
Contour Profi le r
36
35
<20%
CO2 ( %)
Feed (X)
6.7
1
35
Curr ent X
Temperature (C)
.
11
40
.
pH
6.8
0
34
aFucosylation
Factor
DO (% )
6.9
For the production bioreactor the limits of
Design Space are defined by a subset of
CQAs:
Galactosylation
aFucosylation
Hi Limit
.
5.9
30.9
674274.3
19000
A ggregates ( %)
Temperature (C)
Curr ent Y
15000000
< 2%
.
11
0
19
Conto ur
Titer ( g/L)
pH
Hi Limit
CEX % A cidic V ariants
12
>40%
Galac tosylation (%)
6.6
1966.2
aFucosylation
360
IVCC (e6 cells/mL)
40
.
15000000
DNA (ppm)
1.2
Feed (X)
.
11
694855.9
19000
HCP ( ppm)
100
Cultur e Duration ( days)
6.6
Response
6.7
20
15000000
Galac tosylation (%)
15000000
6.85
6.8 [Medium] (X)
Osmo lality (mOs m)
6.8
2
25.7
Titer ( g/L)
50
pH
pH
34
35
Hi Limit
.
5.2
20
Contour ProfiGalac
le rtosylation (%)
Curr ent X
6.9 CO2 ( %)
V ert
6.9
Lo Limit
4.6
2
7.1
V ert
DO (% )
Horiz
36
Feed (X)
Response
.
2
Contour
7.1 Profi le r
6.7
Lo Limit
Cultur e Duration ( days)
34
34.5
Lo Limit
6.6
Horiz
Curr ent Y
0
35.5
aF ucosylation
<20%
5.0
.
Factor
6.5
2
Temperature (C)
29.8
20
DO (% )
697946.1
.
6.8
CO2 ( %)
2040.3
.
pH
30.2
20
6.7
[Medium] (X)
1.5
.
Osmo lality (mOs m)
15000000
7.5
6.7
40
A ggregates ( %)
Conto ur
Galac tosylation (%)
V6.9
ert
40
5.7
19000
1
17
>40%
IVCC (e6 cells/mL)
6.9
pH
0
aFucosylation
Galac tosylation 160
(%)
0
Horiz
2
6.6
17
7
Curr ent Y
36
Temperature (C)
Titer ( g/L)
12
IVCC (e6 cells/mL)
7
Conto ur
Curr ent
Y le r
Contour
Profi
440mOsm
12
7.1
1
IVCC (e6 cells/mL)
440
17
Osmolality
Response
Titer ( g/L)
Curr ent X
aFucosylation
360mOsm
400mOsm
35
Galac tosylation (%)
50
HCP ( ppm)
40
DNA (ppm)
6.85
CEX % A cidic V ariants
1.2
A ggregates ( %)
360
Factor
pH
V ert
6.9
Intersection of all CQA models define the
Design Space
1.2
Gal ac tosylati on (%)
Feed (X)
12
7.1
1
IVCC (e6 cells/mL)
7
Contour
Profi le r
Horiz
50
100
6.85
[Medium] (X)
Titer ( g/L)
Curr ent X
34
34.5
35
aFucosylation
35
Temperature (C)
Galac tosylation (%)
50
HCP ( ppm)
70
DNA (ppm)
6.85
CEX % A cidic V ariants
1.2
A ggregates ( %)
400
.
20
Design Space for Culture Duration 17 Days
DNA (ppm)
35
CO2 ( %)
Cultur e Duration ( days)
6.6
Response
17
Contour Profi le r
Conto ur
Curr ent X
<20%
Temperature (C)
Osmo lality (mOs m)
360
Response
Factor
pH
1.2
>11%
Osmo lality (mOs m)
.
31.0
3
DO (% )
6.85
[Medium] (X)
V ert
7
6.9
100
aF ucosylati on
pH
6.8
34
Curr ent X
Temperature (C)
.
Contour
7.1 Profi le r
Gal ac tosylati on (%)
Factor
pH
V ert
7
6.9
40
15000000
40
A ggregates ( %)
7.1
Contour
Profi le r
11
.
6.6
Horiz
.
2
20
458789.8
19000
Hi Limit
.
6.6
34.4
15000000
DNA (ppm)
Lo Limit
4.5
2
40
HCP ( ppm)
3
Curr ent Y
0
11
.
>40%
One model for each CQA: describes
relationships with CPPs
12
Titer ( g/L)
15000000
A better way to look at the data:
.
11
Feed (X)
Response
.
9.8
6.7
40
A ggregates ( %)
Hi Limit
.
36.8
6.8
469303.1
19000
40 mmHg
.
40
Hi Limit
Cultur e Duration ( days)
34
34.5
35
Lo Limit
5.2
6.9
pH
100 mmHg
HCP ( ppm)
.
IVCC (e6 cells/mL)
7
Curr ent Y
11
160 mmHg
Galac tosylation (%)
DNA (ppm)
19000
6.6
0
CO2
40
15000000
40
1498.3
< 2%
Lo Limit
<20%
4.7
.
Factor
7.8
2
Temperature (C)
33.7
20
DO (% )
495754.0
.
6.8
CO2 ( %)
1552.2
.
pH
30.2
20
6.7
[Medium] (X)
1.2
.
Osmo lality (mOs m)
V ert
6.9
15
Conto ur
Titer ( g/L)
aFucosylation
0
Horiz
2
12
7.1
1
IVCC (e6 cells/mL)
Response
440mOsm
.
11
15000000
aF ucosylati on
7 ent Y
Conto ur
Curr
Contour
Profi le r
Titer ( g/L)
360mOsm
400mOsm
Curr
ent X
aFucosylation
35
Galac tosylation (%)
50
HCP ( ppm)
40
DNA (ppm)
6.85
CEX % A cidic V ariants
1.2
A ggregates ( %)
360
pH
pH
Factor
15
Osmolality
7
V6.9
ert
2
20
.
A
ggregates ( %)
400
7.1
1
Hi Limit
.
5.7
30.3
490873.2
19000
CEX
1.2 % A cidic V ariants
3
Feed (X)
Lo Limit
4.2
2
Galac
50 tosylation (%)
15000000
Curr ent Y
0
aFucosylation
35
Gal ac tosylati on (%)
Contour Profi le r
Horiz
1
Cultur e Duration ( days)
Lo Limit
4.8
.
V ert Factor
8.3
2
Temperature (C)
33.4
20
DO (% )
513494.5
.
CO2 ( %)
1471.7
.
pH
28.2
20
[Medium] (X)
1.3
.
Osmo lality (mOs m)
Design Space for Culture Duration 15 Days
CEX % A cidic V ariants
12
IVCC (e6 cells/mL)
15
Conto ur
Curr ent
Y le r
Contour
Profi
0
Horiz
11
440
Feed (X)
1
Cultur e Duration ( days)
Titer ( g/L)
1.2
Osmo lality (mOs m)
12
Response
6.85
[Medium] (X)
360
Feed (X)
100
pH
6.6
34
34.5
35
35.5
36
Temperature (C)
Vinci/Defelippis - CMC BWG QbD
34
34.5
35
35.5
36
Temperature (C)
Lilly - Company Confidential 2010
18
Contour Profiler
Design Space Based on Process Capability
Horiz Vert Factor
Temperature (C)
DO (%)
CO2 (mmHg)
pH
[Medium] (X)
Osmo (mOsm)
Feed (X)
IVCC (e6 cells/mL)
Duration (d)
Current X
35
50
40
6.85
1.2
360
12
1
15
Understanding Variability
Example: Day 15, Osmo=360 mOsm
and
pCO2=40 mmHg Contour Current Y Lo Lim it
Response
Titer (g/L)
aFucosylation
Galactosylation (%)
HCP (ppm)
DNA (ppm)
CEX % Acidic Variants
3
11
40
675000
2250
40
5.3408326
9.1879682
38.227972
466955.66
1382.1644
34.420095
3
.
.
.
.
.
>99%
confidence of
satisfying all
CQAs
Hi Lim it
.
11
40
.
.
.
50% contour
approximates “white”
region” in contour plot
0. 99
7.1
7.05
7
7
9
0. 0.80.7 0.5
aFucos >11%
0.
95
6.95
pH
pH
6.8
Galactosylation (%)
0.99
95
0.
pH
5
0.2
6.9
6.9
0.
99
0.
9
0.
8
0.7
6.85
0.99
0.5
6.8
0.25
aFucosylation
Galact >40%
0.8
0.7
0.5
0.25
6.75
6.7
0.95
0.9
0.95
0.9
0.8
0.7
0.5
0.25
6.7
6.65
6.6
34
34.5
35
35.5
36
6.6
34
Temperature (C)
Temperature (C)
Vinci/Defelippis - CMC BWG QbD
Case Study
34.2
34.4
34.6
34.8
35
35.2
35.4
35.6
35.8
36
Temperature (C)
Lilly - Company Confidential 2010
19
Risk Assessment Approach
Multiple Assessments Throughout the
A-Mab Development Lifecycle for Entire Process
Process 2
Quality
Attributes
Process 1 2
Life Cycle
Management
Design Space
Prior Knowledge
Process Understanding
Process
Development
Process
Characterization
Product Understanding
Draft Control
Strategy
Process
Performance
Verification
Final Control
Strategy
Process
Parameters
Risk
Assessment
Risk
Assessment
Risk
Assessment
Risk
Assessment
You Are Here
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
20
Control Strategy for Upstream Production
Quality-linked
Process Parameters
(WC-CPPs)
Key Process
Parameters
(KPPs)
Temperature
Time
Temperature
pH
Dissolved CO2
Culture Duration
Osmolality
Remnant Glucose
Vinci/Defelippis - CMC BWG QbD
Case Study
Working Cell Bank
Step 1
Seed Culture Expansion
in Disposable Shake
Flasks and/or bags
Viable Cell Concentration
Viability
Temperature
pH
Dissolved Oxygen
Culture Duration
Initial VCC/Split Ratio
Step 2
Seed Culture Expansion
in Fixed Stirred Tank
Bioreactors
Viable Cell Concentration
Viability
Antifoam Concentration
Time of Nutrient Feed
Volume of Nutrient Feed
Time of Glucose Feed
Volume of Glucose Feed
Dissolved Oxygen
Step 3
Production Culture
Step 4
Centrifugation and Depth
Filtration
Clarified Bulk
In-Process
Quality Attributes
Viable Cell Concentration
Viability
Temperature
Culture Duration
Initial VCC/Split Ratio
Flow Rate
Pressure
Controlled within the
Design Space to
ensure consistent
product quality and
process performance
Key Process
Attributes
Product Yield
Viability at Harvest
Turbity at Harvest
Bioburden
MMV
Mycoplama
Adventitious Virus
Product Yield
Turbidity
Controlled within acceptable
limits to ensure consistent
process performance
Lilly - Company Confidential 2010
Assay results part
of batch release
specifications
Slide 21
Example of Control Strategy for Selected CQAs
CQA
Criticality
Process
Capability
Testing
Criteria
Other Control
Elements
Aggregate
High (48)
High Risk
DS and DP
release
Yes
Parametric Control of
DS/DP steps
aFucosylation
High (60)
Low Risk
DS Process
Monitoring
Yes
Parametric Control of
Production BioRx
Galactosylation
High (48)
Low Risk
DS Process
Monitoring
Yes
Parametric Control of
Production BioRx
High (24)
Very Low
Risk
Charact.
Comparability
Yes
Parametric Control of
Prod BioRx, ProA, pH
inact, CEX , AEX steps
DNA
High (24)
Very Low
Risk
Charact.
Comparability
Yes
Parametric Control of
Prod Biox and AEX
Steps
Deamidated
Isoforms
Low (12)
Low Risk
Charact.
Comparability
No
Parametric Control of
Production BioRx
Host Cell
Protein
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
22
Lifecycle Management of Design Space
Dynamic Modeling
Challenge:
• Data from a limited number of batches is required for process validation
ex: n=5 or more for 3 bioreactors ; costly and often critical path
• Limited replicates are not statistically significant – at best test the “system”
including facility, equipment, process, operators, etc
Alternative Lifecycle Approach or Continuous Process Verification:
• Quality Mgt System assures site’s readiness and compliance
• Use 1 or 2 batches to confirm or demonstrate validity of design space
• Utilize a multivariate statistical partial least squares (PLS) model for continuous
process verification as commercial experience grows in number of runs
• Scheduled reviews of product quality data trends and design space validity during
the product lifecycle
Vinci/Defelippis - CMC BWG QbD
Case Study
Lilly - Company Confidential 2010
23
Design Space Linkage to Critical Attributes
Successful acceptance or utilization of our evolving view of design space relies
on linking the multiple elements of documented knowledge and systems:
Facilitated formal attribute rankings and parameter risk assessments to
guide DOEs
Linkage of critical attributes and parameter ranges used
Delineation of how lifecycle oversight (control strategy) of critical and noncritical parameters and specification/limit testing occurs
Movement to best practices for engineering first principles/mechanistic
models and statistical modeling as they apply to QbD paradigm
Vinci/Defelippis - CMC BWG QbD Case
Study
Lilly - Company Confidential 2010
24
Upstream Development Team
Ilse Blumentals
Guillermo Miroquesada
Kripa Ram
Ron Taticek
Victor Vinci
GSK
MedImmune
MedImmune
Genentech
Lilly
Special thanks to Mike DeFelippis
*Help from many others – CMC BWG member company reps and internal
resources at each company
Vinci/Defelippis - CMC BWG
QbD Case Study
Lilly - Company Confidential 2010
25

similar documents