In this session from the Institute of Validation Technology's Validation Week Europe, Kurtis Epp and John Kandl discuss how to implement QbD to all three stages of process validation.
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Applying QbD to Biotech Process Validation
1. BIBioTech
BI
Applying
Validation
A case stu
Stage 1 –
TECH
LOGIC
BioTechLogic, Inc.
BIO TECH
LOGICBIO ®
TECH
IOLOGIC
Logic, Inc.
IO ®
g QbD to Biotech Process
n :
udy in applying QbD to
– Process Design
IVT Validation Week
28-30 March 201128 30 March 2011
Kurtis Epp, John Kandl
BioTechLogic, Inc.
2. Agenda
• Process Backgrou
• Risk Assessment
• DOE
• Parameter Evaluat
• Conclusions
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
und
tion
Slide 2
Company Confidential
3. Process Back
The scope of this case stup
data obtained from small-
Chromatography runs per
limits in experiments planlimits in experiments plan
Design of Experiments (D
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
kground
udy is to provide and evaluatey p
-scale Ion Exchange
rformed within defined parameter
ned and executed according toned and executed according to
DOE).
Slide 3
Company Confidential
5. Risk Assessm
Operating Parameter Normal Operating
UV at Start of AEX Fraction Collection > 0.50 AU
UV at End of AEX Fraction Collection < 20% of maximum p
Elution Gradient
8% per CV
Elution Gradient
(20 – 100% B ove
Equilibration volume ≥ 5 CV
Load flow rate 100 – 120 L/h (Targe
Wash flow rate 100 – 120 L/h (Targe
Elution flow rate 100 – 120 L/h (Targe
Pre-Equilibration flow rate 100 – 120 L/h (Targe
F1 F3: 0 2
Fraction Volume
F1 – F3: 0.2
F4+: 0.1 CV
Wash volume ≥2 CV
Fraction Mixing Speed 65 rpm
Fraction Mixing Time 20 - 30 min
Pool Mixing Time
10 – 15 min
Pool Mixing Time
Pool Mixing Speed
100 rpm
Pre-Equilibration volume ≥ 4 CV
WFI Rinse volume ≥ 3 CV
WFI Rinse flow rate 100 – 120 L/h (Targe
Equilibration flow rate 100 – 120 L/h (Targeq ( g
In-Process Control Limits
Column Load 15 – 25 g / L r
Column Bed Height 30 ± 3 cm
Column Backpressure during
Equilibration, Load, Wash, and Elution
< 3 bar
Effl t H t d f E ilib ti
± 0.3 pH Units of Eq
Effluent pH at end of Equilibration
p q
Buffer pH
Effluent Cond. at end of Equilibration
± 1 mS/cm Units of E
Buffer Conduc
Effluent UV at the end of Equilibration Zero
Fraction Pooling Criteria (RP-HPLC) ≥ 95% Main P
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
ment
g Range Potential Effect(s) of Failure SEV OCC DET RPN
U Product loss, quality 9 4 2 72
peak height Product loss, quality 9 4 2 72
V
Inconsistent quality 8 2 4 64
r 10 CV)
Inconsistent quality 8 2 4 64
Product loss, quality 9 2 1 18
et: 110 L/h) Inconsistent quality 4 1 3 12
et: 110 L/h) Inconsistent quality 4 1 3 12
et: 110 L/h) Inconsistent quality 4 1 3 12
et: 110 L/h) Longer equilibration 4 1 3 12
CVCV
V
Product loss, quality 9 1 1 9
Inconsistent quality 5 1 1 5
Inconsistent quality 4 1 1 4
n Inconsistent quality 4 1 1 4
n
Non-homogeneity, inconsistent
sampling/yield
4 1 1 4
sampling/yield
Non-homogeneity, inconsistent
sampling/yield
4 1 1 4
Longer equilibration 3 1 1 3
Longer equilibration 3 1 1 3
et: 110 L/h) Longer equilibration 1 1 1 3
et: 110 L/h) Product loss, quality 3 1 1 3) , q y
resin
m
quilibrationq
Equilibration
ctivity
Peak
Slide 5
Company Confidential
6. Risk Assessm
The parameters selectedp
the UV at start of fraction
collection, gradient slope
D d t i blDependent variables ana
and purity by RP-HPLC.
Testin
Input parameter
Lower
UV at the beginning of
fraction collection (AU)
0.40
UV at the end of pooling
(% from peak max)
10
Gradient slope (% per CV) 4.0
Resin Load, (g/L resin) 15
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
ment
d for evaluation in this study werey
n collection, UV at end of fraction
e, and resin load.
l d t i i ldalyzed were protein recovery yield
ng limitsg
Output parameters
Upper
0.60
Purity by RP-HPLC (≥ 95%)
Step yield (≥ 75%)
30
12.0
25
Slide 6
Company Confidential
7. DOE
Pattern Exp. No.:
UV at the sta
of fraction
collection
p
collection
(AU)
−−−− 1 0.4
0000 2 0.5
0000 3 0 50000 3 0.5
−++− 4 0.4
0000 5 0.5
++−− 6 0.6
7 0 6+−−+ 7 0.6
++++ 8 0.6
+−+− 9 0.6
−−++ 10 0.4
0000 11 0.5
−+−+ 12 0.4
Note: It is important to randomize
independence of your observation
mistake by the operator.
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
art
n
UV at the end
of pooling
Gradient
slope
Resin Load
(g/L)
(% peak max) (% per CV)
(g/L)
10 4 15
20 8 20
20 8 2020 8 20
30 12 15
20 8 20
30 4 15
10 4 2510 4 25
30 12 25
10 12 15
10 12 25
20 8 20
30 4 25
e the run order to assure the
ns and reduce the chances of a
Slide 7
Company Confidential
8. Experimental
• A summary of the var
Ion Exchange Chrom
Outputs step yield anOutputs, step yield an
listed in the following
• Runs that did not mee
are shaded to indicate
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
Results
riable parameters for each of the
atography runs as well as the
nd eluate purity by RP HPLC arend eluate purity by RP-HPLC are
table.
et the defined acceptance criteriap
e run failure
Slide 8
Company Confidential
10. Statistical dat
• For evaluation of statisti
on protein quality and q
software package was u
Squares model for EffecSquares model for Effec
• The model was run sep
of yield and purity by RP
• The fractional factorial m
inclusion of all single fac
interactions as model efinteractions as model ef
resource and time const
was not possible.
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
ta analysis
ically significant effects of factors
uantity, the JMP statistical
used applying the Standard Least
ct Leveragect Leverage.
arately for two output parameters
P-HPLC.
model was initially run with the
ctor and some two-factor
ffects (Resolution IV) Due toffects (Resolution IV). Due to
traints, a higher resolution study
Slide 10
Company Confidential
11. Statistical Model P
Step Yieldp
• The final model used for
effects and two factor int
• There were two significa
i ifi t i t ti Tsignificant interaction. T
graphically represented i
plots and interaction conp
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
Parameter Estimates for
assessment all four primaryp y
teractions listed above.
nt main effects and one
h l ihese conclusions were
in the main effects leverage
tour plots.p
Slide 11
Company Confidential
12. Main Effect Lever
a. UV at Start Collection
c. Gradient Slope
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
rage Plots for Step Yield
b. UV at End Collection
d. Resin Load
Slide 12
Company Confidential
13. Statistical Model P
Purity by RP-HPLy y
• The final model used for
primary effects (excludin
two-factor interactionstwo factor interactions.
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
Parameter Estimates for
C
r assessment included all four
ng resin load) as well as three
Slide 13
Company Confidential
14. Two-Factor Intera
Purity by RP-HPLy y
Step Yield (%)
Gradient Slope (%/CV)
0
UV @ End (%)
a. UV at End Collection x Gradient Slope
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
action Contour Plots for
C
Step Yield (%)
Resin Load (g/L resin)
0
UV @ Start (AU)
b. UV at Start Collection x Gradient Slope
Slide 14
Company Confidential
15. Confirmation
• There is one area of th
represents process failp p
yield, run 8. In order to
acceptance criterion is
operated within its defioperated within its defi
modified and two addit
augment the initial des
• We chose to tighten th
as it was determined to
significant primary effesignificant primary effe
analysis.
• None of the experimenp
failed the acceptance c
confirms the predicted
study
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
study.
of Results
he tested design space that
lure space with regard to stepp g p
o ensure that the step yield
always met when the process is
ned PARs the design space wasned PARs, the design space was
tional runs were performed to
sign.
e gradient slope upper limit to 11%
o be easy to control and the most
ect for step yield in our initialect for step yield in our initial
ntal runs for the tightened limitsg
criterion for step yield. This
design space from the initial
Slide 15
Company Confidential
16. Design Space
SpaceSpace
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
e vs Defined Control
Based on the models generated for
h f th d fi deach of the defined process
outputs, a three-dimensional plot
was generated to graphically show
the process control space with
d h i llregard to the experimentally
defined design space
Slide 16
Company Confidential
17. Conclusions
• The results of this DO
design space from the
• Based on our findings
h f th thwe chose for the three
appropriate with the ex
• As a result we tighteneAs a result we tightene
• The IPC limits for resin
set appropriately.
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
E study confirm the predictedy p
e initial study.
we determined that the IPC limits
i t te input parameters were
xception of Gradient Slope.
ed our control limitsed our control limits.
n load were also determined to be
Slide 17
Company Confidential
18. Final Paramet
Normal Ope
Operating Parameter
Normal Ope
(NO
UV at the beginning of
pooling, % from peak max Target: >p g p
UV at the end of pooling, %
from peak max Targe
Gradient slope, % per CV TargeGradient slope, % per CV Targe
Performance Parameter In
Resin Load, g/L
* Note: The claimed experimentally confirmed range is slightly tighter than th
support the higher end of the range.
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
ter Limits
rating Range Experimentally Confirmedrating Range
OR)
Experimentally Confirmed
Ranges
> 0.5 AU 0.4 – 0.6 AU
t: 20% 10 – 30%
et: 8% 4 – 11%*et: 8% 4 11%
n-Process Acceptance Criterion
15 – 25 g/L
he design space in that it takes the worst-case data point to
Slide 18
Company Confidential
19. Thank You
BioTechLogic Inc serBioTechLogic Inc. ser
biopharmaceutical ind
with companies to mwith companies to m
resource and comme
TECH
LOGICBIO TECH
LOGICBIO ®BioTechLogic, Inc.
rves therves the
dustry by collaborating
eet their developmenteet their development,
ercialization needs.
www.biotechlogic.com
www.processvalidation.com
Slide 19
Company Confidential