- The document discusses the development of potency assays using the Octet platform for biotherapeutics. - Case studies are presented showing potency assays can be developed in 5-6 days using Octet compared to 5-7 days using other technologies like ELISA or SPR. - Assays developed on Octet show good accuracy, precision, specificity, linearity and range meeting regulatory guidelines. - Potency assays have been used to support process development and manufacturing through to quality control.

1. Nathan Oien, Ph.D.
Senior Scientist, Analytical Development
ForteBio User’s Meeting 2017

2. • About KBI: A Contract Development and Manufacturing
Company
• KBI helps partners accelerate and optimize drug
development and manufacturing programs with an
extensive suite of expert development and manufacturing
services in an agile, client-friendly partnering environment
2

3. • Development
• Method Transfer
• Method Development
• Method Optimization
• PD support (CLD, Upstream
and Downstream)
• Lots of Samples!
• Quality Control
• Method Qualification
• Stricter Requirements
• Fewer Samples
3
Challenge: Technology that can be utilized in a Q.C.
and in a high-throughput fashion
Answer: Platform Method allows for rapid
development, optimization, and PD Support

4. • In-Scope
• Platform Potency Assay Development
• Use of potency assays throughout process development and GMP
testing
» Four case studies will be reviewed
• Out-of-Scope
• Kit based ELISAs (HCP, ProA)
• Octet used as a HT screening tool

5. • ICH Q6B:
• Must assess biological properties, ligand based assays may
acceptable
• The potency assay must measure biologically relevant activity
• Biological assay must have a calibrated reference, which can
included in-house reference material

6. • ICH Q2 R1 Guidelines
• Assays should be developed to pass Accuracy, Precision,
Specificity, Linearity, and Range.
• Define Conditions
• Accuracy – Actual versus predicted relative potency is 80 – 120%
• Precision – Replicate Runs and different instruments
• Specificity – Binding to similar proteins
• Linearity – 50 to 150% of the nominal range evaluated
• Range – Evaluated across linearity range

7. • ELISAs
• Decide on format (sandwich, single antibody, competitive)
• Evaluate ligand loading (2 Days)
• Evaluate incubation times (1 Day)
• Evaluate range (1 Day)
• Fine tune the assay (1 Day)
• Total development time is 5 days
• SPR
• Chip selection (1 Day)
• Ligand chemistry and density selection (1 Day)
• Buffer optimization (1 Day)
• Range evaluation (1/2 Day)
• Flow rate optimization and analyte concentration determination (1
Day)
• Development of wash conditions (1 Day)
• Total development time is 5.5 days

8. • BLI
• Chemistry selection (1/2 Day)
• Ligand loading evaluation (1/2 Day)
• Buffer Evaluation (1/2 Day)
• Final optimization (1/2 Day) including Sample dilution
• Prior to proceeding with development consider the protein
• What is the binding target?
• How large is the protein of interest and the binding target?
• Evaluate commercial source of the binding target (Abcam, Sino,
and R&D systems), including modified proteins (HIS, Biotinylated,
etc).
• What is the anticipated binding affinity?

9. • Sensor tips and chemistries should be selected with the
end in mind
Amine
Reactive
Super
StrepStrepNi-NTAAnti-FCLoading Affinity
Cost
Robustness*
Protein Availability
Super
Strep
StrepAnti-FC Ni-NTA
Amine
Reactive
Anti-FC Ni-NTA
Super
Strep
Strep
Amine
Reactive
Amine
Reactive
Super
Strep StrepNi-NTAAnti-FC
*Extra weight is added to robustness
Overall Score Ni-NTAAnti-FC
Super
Strep Strep
Amine
Reactive

10. 10
Baseline
30 – 60 sec.
Loading
~ 3-5 minutes
Baseline/
Wash
60 – 120 sec.
Association
2 – 10+ minutes
Y Y
Platform: 60 sec. 5 minutes 60 sec. 5 minutes

11. Load Density Scouting
• Sensor loading is evaluated using an antigen/protein to ensure
sufficient protein is loaded on the sensor, but not overloaded.
11

12. Concentration scouting of 2nd molecule (antibody)
• Holding the first molecule (antigen) constant, a concentration
range is evaluated for product association in a concentration
dependent manor.
12
For typical antibodies, assay development would be done => method
optimization and assessment (accuracy, linearity, etc.)
% 𝑅𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑃𝑜𝑡𝑒𝑛𝑐𝑦
𝑂𝑐𝑡𝑒𝑡 𝐺𝑒𝑛𝑒𝑟𝑎𝑡𝑒𝑑 𝐶𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛
𝑇ℎ𝑒𝑜𝑟𝑒𝑡𝑖𝑐𝑎𝑙 𝐶𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛

13. • Qualified 6 single target potency methods using different
antibody/antigen pairs
• Frequently achieve ≤10% RSD for sample triplicates and 90 – 110%
Potency
• Release/stability method for activity much faster than typical ELISA or
SPR
13
R2 ≥ 0.99

14. • Case Study #1:
• IgG1 method development Fc binding Development: Evaluating
the importance of wash and sample buffers and plates
• Case Study #2:
• Fusion Protein (non-mAb) Titer Development – PD support only
• Case Study #3:
• Antigen potency development for vaccine using full curves
• Case Study #4:
• Following potency assay data through QC and preparing for
validation

15. • Method Development:
• Investigation of addition of excipient to wash and/or sample
diluent solutions
Assay Excipient
Wash
Non-
Excipient
Wash
Excipient Std
Diluent
Non-
Excipient Std
Diluent
1 X X
2 X X
3 X X
4 X X

16. Excipient Wash
No Excipient Sample Diluent
Excipient Wash
Excipient Sample Diluent
No Excipient Wash
Excipient Sample Diluent
Non-Excipient Wash
Non-Excipient Sample Diluent

17. Half Area PlatesFull Area Plates
Noise
Baseline drop
Wider variation

18. • Wash and Loading buffers should be prepared without the
excipient addition
• Plate type is important, similar to a fluorescence assay
• Method was successfully developed an implemented into a
QC assay 400
200
50
25
Blank
100
Loading

19. • Objective: Develop a kinetic binding curve to determine KD
values for a protein of interest.
Work done in collaboration with The Haynes Lab at Duke,
DAIDS, and ABL.

20. Initial conditions tested
Extend loading step
Decreased loading step, increase concentration

21. Final conditions demonstrate a clear dose response increase
concentration

23. Octet® Method Biacore® Method
Sample KD (nM)
Fold difference
from Control Sample KD (nM)
Fold difference from
Control
Octet Reference Material 18.2 NA Biacore Reference Material 26.0 NA
Process 1 36.7 2.0 Process 1 94.9 3.7
Process 2 27.5 1.5 Process 2 41.8 1.6
Process 3 20.0 1.1 Process 3 36.9 1.4
Process 4 17.2 0.9 Process 4 24.8 1.0
Process 5 21.4 1.2 Process 5 50.5 1.9

24. • KD values and/or dose response curves can be used as an
alternative approach to concentration based potency
• KD values are comparable to SPR data
• Robust assays can be rapidly developed and used to
support in-process analysis

25. • Protein Titer Development using anti-FC antibody Sensors
YYYY YYYYY YYYYY

26. Sample
Well Conc.
(µg/mL) Dilution
Calc. Conc
(µg/mL)
HPLC Titer
results
(µg/mL)
Transient 2.18 10 21.80 73.00
Stable 21.60 10 216.00 225.00
117-1 0.00 10 <15.6 40.00
117-2 0.00 10 <15.6 27.00
117-3 0.00 10 <15.6 31.00
117-4 0.00 10 <15.6 20.00
117-5 2.97 10 29.70 47.00
117-6 0.00 10 <15.6 22.00
117-7 0.00 10 <15.6 29.00
117-8 0.00 10 <15.6 15.00
r2 = 0.9963
Chi Square = 0.0001
Dilutional Linearity indicated a 10 fold
dilution was needed

27. Results of analysis of single plate for titer
Data in this range was suspiciously high

28. 2 3 4 5 6 7 8 9 10 11 12
A 35.8 77.9 85.7 29.5 43.1 51.3 47.4 48 55.9 30 25.3
B 19.5 35 52 18.7 23.4 29.2 26.4 29.1 33.6 18.5 41.8
C 11.6 29.1 34.1 12.9 12.6 16.3 16 21.3 24.3 24.2 24.8
D 6.09 17.9 12 8.01 6.33 8 10.2 11.9 10 11.8 10.7
E 4.36 8.29 8.7 4.3 5.69 6.96 5.74 8.95 8.2 9.07 8.51
F 3.13 6.82 4.13 4.56 2.75 6.07 2.94 5.14 5.17 6.28 4.98
G 2.15 3.59 3.89 3.22 2.71 3.94 3.23 3.15 4.36 5.29 10.8
H 1.61 2.61 2.76 1.84 1.97 3.02 2.42 2.9 3.1 3.66 2.42
5 plates were analyzed each plate had the same pattern as displayed below

29. Condition 1
Measured
Concentration
(µg/ml)
DF Corrected
Concentration (µg/ml) Spiked result %Spike Recovery
Sample 1 2.74 54.8 6.06 83%
Sample 2 2.56 51.2 6.34 95%
Sample 3 2.32 46.4 7.41 127%
Sample 4 3.85 77 9.21 134%
Condition 2
Measured
Concentration
(µg/ml)
DF Corrected
Concentration (µg/ml) Spiked result %Spike Recovery
Media Spike 0 0 4.4 110%
Sample 2 1.9 38 5.86 99%
Sample 3 2.35 47 43.8 1036%
Sample 4 4.57 91.4 26 536%
Two lots of samples were analyzed from different runs using different media
conditions

30. • Always scout dilutional linearity if evaluating samples in
cell culture harvest
• Spike recovery should always be preformed with all
matrixes at the correct dilution
• Follow the data through process development and
evaluate the data in the plate layout

31. • Potency Assay followed our platform approach
• Assay developed to measure binding of an antibody to an
antigen
• Antigen immobilized using Ni(NTA) sensors
• Standard curve ranged from 400% to 12.5% of nominal
load
• Potency measured based on nominal load comparison

32. Target Mean STD DEV %RSD %Recovery
400 400 21 5 100
200 200 4 2 100
100 99 1 1 99
50 51 1 2 101
25 25 0 1 100
12.5 12.5 0 3 100
0 N/A N/A N/A N/A
Linearity
Target Avg Binding Rate Average Potency %RSD
Analyst 1 100 0.7 96
Analyst 2 100 0.6 103
6
Intermediate Precision

33. • Standard curve R2 ≥0.97 (Range observed 0.97 to
0.999)
• %RSD of triplicate standard preparations ≤20% (
Range observed 0 to 11%)
• Recovery of standards must be between 70 – 130%
(range observed 89 to 110%)
• Blank must read “too low” (sometimes a value is
registered)

34. Binding rate of
nominal load over
1 year period

35. • Standard curve R2 ≥0.97 to ≥0.98
• %RSD of triplicate standard preparations ≤20% to ≤15%
• Recovery of standards must be between 70 – 130% to 80 –
120%
• Blank must read “too low” or binding rate less than a set
limit (I.E. ≤0.05 nm/min)
• Add a control sample
• Set binding rate range on standards

36. • Carson Cameron
• Brendan Peacor, Ph.D.
• Jimmy Smedley, Ph.D.
• Abhinav Shukla, Ph.D.
• Melissa Ragan
• Amanda Hoertz, Ph.D.
• KBI Biopharma Analytical Development Team
• ForteBio
• Hope Boyce
• Hendrick Loei
• David Apiyo, Ph.D.
• Yuichi Sugiyama