2. 2
• Program overview
• Background
• Scope of Work
• Virus Filters for Evaluation
• Model Molecules
• Experimental Plan & Results
3. 3
• Our current preferred viral filter is the Asahi Kasei Planova® 15N filter as it
consistently provides excellent log reduction value (LRV) of small viruses at ~
20 nm
• Limitations include lower batch throughput for feeds with higher product
concentrations, the need for a gold particle test for integrity testing, and high
operational cost
Molecule Type Viral Filter
Manufacturing Target
Loading (L/m2
)
XMuLV LRV
(80-110 nm)
MMV LRV
(18-24 nm)
IgG4 Planova® 15N ≤1152
6.42 3.94
IgG4 Planova® 15N ≤351
≥ 6.08 3.80
gp120 Planova® 15N ≤501
4.72 4.99
Recombinant Enzyme Planova® 15N ≤1002
5.34 4.08
IgG1 Planova® 15N ≤1002
3.93 4.48
1 Established as true limits during PD development;
2 Limit based on material availability for the PD development study
XMuLV = Xenotropic Murine Leukemia Virus; MMV = Mouse Minute Virus
4. 4
• As a CMO focusing on clinical development of > 12 IND stage
molecules per year, we are looking for a viral filter that offers
consistent and reliable performance from product to product
• The filter must meet the following requirements:
• High capacity and scalable
• Disposable flow path at manufacturing scale
• Easy installation, utilization, and integrity test
• Consistent and robust clearance of parvovirus
• There are several established virus filters that are currently used in
industry including the Asahi Kasei Planova®, EMD Millipore Viresolve®,
and Sartorius Stedim Virosart®
• Sartorius Stedim recently launched a hollow fiber Virosart® filter in March 2015
• EMD Millipore has a new Shield H filter that will be launched this year (beta
version tested in this study)
5. • EMD Millipore and Sartorius Stedim virus filters can be integrity tested
by diffusional flow rate of air through the water wetted membrane
• The air/water diffusion integrity test is currently standard practice for
other filters in our manufacturing facility
• With the exception of Planova® BioEX, the Asahi Kasei filters require
the use of the gold particle test for integrity testing
• The GPT could be eliminated if the Planova® BioEX, EMD Millipore or
Sartorius Stedim viral filters were integrated into the platform
• The elimination of the gold particle integrity test would help simplify
and reduce the number of operations in manufacturing
5
6. • Identify a platform filter that offers increased batch throughput,
consistent performance from product to product, and allows for
easy manufacturing operations
• The filter could become part of our platform approach to
monoclonal antibody (mAb) purification
• Rely on the vendor for viral clearance data to demonstrate
acceptable LRV for each filter type being evaluated
6
7. 7
• Rely on LRV data provided by vendors
• All filters demonstrate similar clearance of small viruses (18-24 nm)
Pre-Filter Viral Filter Vendor Viral Filter Composition MMV LRV PPV LRV
0.1 µm PES
(VPF, Shield H
beta evaluated)
Viresolve® Pro
(Vpro)
EMD Millipore PES Asymmetric Membrane
5.8
N/A4.8
5.7
0.1 µm PES Planova® 15N
Asahi Kasei
Bioprocess
Regenerated Cellulose Hollow
Fiber
N/A > 4.6
0.1 µm PES Planova® 20N
Asahi Kasei
Bioprocess
Regenerated Cellulose Hollow
Fiber
6.9
> 5.2
> 5.1
> 4.2
0.1 µm PES Planova® BioEX
Asahi Kasei
Bioprocess
PVDF Hollow Fiber
> 4.8
> 5.3
> 6.1
0.1 µm PES Virosart® HF Sartorius Stedim PES Hollow Fiber 6.2 N/A
EMD Millipore Viresolve® Pre-filter (VPF) = diatomaceous earth, positive charge resin, cellulose esters
EMD Millipore Shield H beta Pre-filter = Mixed mode PES
PPV = Porcine Parvovirus
1 LRV data provided by vendor
1 1
8. • Model molecules selected for testing in the various viral filters:
• All material was stored at -80°C so a single freeze-thaw cycle was
required; a 0.1 µm PES filtration was performed just prior to viral
filtration operation to remove potential aggregates generated due to
freeze-thaw
8
Molecule
mAb
Type
Product
Concentration
(mg/mL)
Buffer Matrix
Feed
Conductivity
(mS/cm)
Feed pH
HMW
Levels
by SEC
mAb1 IgG1 3.9 20 mM phosphate, 10 mM citric acid, 300 mM NaCl, pH 5.0 29-31 5.0 2.3%
mAb2 IgG1 5.4 50mM Tris, 240mM NaCl, pH 8.0 25 8.0 3.3%
mAb3 IgG4 11.6 10 mM phosphate, 140 mM NaCl, pH 6.1 13-16 6.1 5.8%
mAb4a IgG4 10.5 10 mM phosphate, 140 mM NaCl, pH 6.1 13-16 6.1 3.1%
mAb4b IgG4 3.3 20 mM phosphate, 0.6M AmSO4, pH 6.1 83-105 6.1 2.3%
9. • Evaluate model molecules in selected virus filter options, which
includes new and established virus filters
• The naked virus filters are evaluated for a direct comparison
• The Viresolve® Prefilter and new Shield-H are evaluated to assess the
performance impact with the inclusion of pre-filtration with the EMD
Millipore virus filter
• Monitor filter performance with the following parameters for each
model molecule:
• Flow decay at a fixed volumetric loading
• Throughput at a fixed volumetric loading
• Operational cost at 25% flow decay for 100L batch volume
» 25% flow decay criteria based on Parenteral Drug Association (PDA)
committee recommendation corresponding to ~ 4 LRV
9
Lute, S., Riordan, W., and Pease, L. et al. (2008) PDA J Pharm Sci and Tech 62, 318-333
11. • The Planova® 20N and BioEX filters demonstrated the lowest flow decay
among the virus filters evaluated
• The Viresolve® Pro and Planova® 15N filters possessed the highest flow
decay
11
N/A
Note: mAb4a was not evaluated in BioEX due to limited material availability
12. • Throughput calculated based on processing time required to achieve a
fixed volumetric loading of 50 L/m2
• The Viresolve® Pro and Virosart® HF filters were the top performers
• The Planova® filters demonstrated the poorest throughput
12
N/A
*
*
* 50 L/m2 loading was not achieved due to significant flux decay therefore throughput per time could not be calculated
Note: mAb4a was not evaluated in BioEX due to limited material availability
13. • Cost calculated based on surface area required to process 100-liter batch
up to 25% flow decay
• The Planova® 20N and BioEX filters were the most cost effective among the
filter options for the molecules tested
• The Viresolve® Pro and Virosart® HF filters possessed the highest
operational cost among the various filters
13
N/A
<
<
<
< <
1, 2 1, 2 2 22
1 Cost includes the gold particle test
2 Cost includes a 0.1µm PES filter
Note: < was included for runs
that did not reach 25% flux decay
therefore the cost was calculated
using the test throughput
15. • The inclusion of the EMD Millipore pre-filters significantly reduced
flow decay compared to the naked Viresolve® Pro filter
• The Viresolve® pre-filter + Pro virus filter (VPF + VPro) and Planova®
20N filter setups demonstrated the lowest flow decay
15
N/A
16. • Throughput calculated based on processing time required to achieve a
fixed volumetric loading of 50 L/m2
• With the exception of mAb2, the inclusion of the EMD Millipore pre-
filters significantly enhanced throughput over time across all molecules
tested in comparison to naked VPro
16
N/A
*
*
* 50 L/m2 loading was not achieved due to significant flux decay therefore throughput per time could not be calculated
Note: mAb4a was not evaluated in BioEX due to limited material availability
17. • Cost calculated based on surface area required to process 100-liter batch
up to 25% flow decay
• The Viresolve® pre-filter (VPF) significantly reduced cost in comparison to
the naked VPro
• The Viresolve® pre-filter + Pro virus filter (VPF + VPro) and Planova® 20N
setups were the most cost effective filters among those evaluated
17
N/A
<
<
<
< <<
Note: < was included for runs that
did not reach 25% flux decay
therefore the cost was calculated
using the test throughput
1 Cost includes the gold particle test
2 Cost includes a 0.1µm PES filter
1, 2 1, 2 2 22
18. 18
• The Planova® filters can be classified as low flux filters with low fouling incidence however the
Planova® 15N fouling incidence is highly dependent on the feed material
• The Viresolve® Pro and Virosart® HF virus filters can be classified as high flux filters with the
Viresolve® Pro filter demonstrating the highest fouling incidence among all virus filters evaluated
• The use of the EMD Millipore pre-filters facilitates achievement of the range of desirability (i.e.
high flux and low fouling incidence)
Note: Data points are average values for all molecules tested
19. • A viral filtration performance study was completed for four monoclonal
antibodies possessing varied product concentrations and buffer matrices
• The Planova® 20N and BioEX filters presented the lowest flow decay,
lowest throughput and lowest cost in comparison to the Viresolve® Pro
and Virosart® HF
• The inclusion of a pre-filter significantly improved the performance and
reduced operational costs of the Viresolve® Pro filter
• The Viresolve® pre-filter + Pro virus filter (VPF + VPro) and Planova® 20N
filter setups demonstrated the lowest flow decay and operational costs
for all molecules in comparison to the other filter setups
• This initial screening study has identified the Millipore Viresolve® pre-
filter + Pro virus filter and the Asahi Kasei Planova® 20N filter setups as
top contenders for integration into the KBI mAb platform
• Additional model molecules will be evaluated with the top performers to
make a final filter section for the platform
19
20. KBI Biopharma, Inc.
Process Development:
• Carnley Norman, Ph.D
• Abhinav Shukla, Ph.D
• Sigma Mostafa, Ph.D
• Leslie Wolfe, Ph.D.
• Aishwarya Sivakumar
Analytical Development:
• Jimmy Smedley, Ph.D.
• Helena Gaweska
• Mike Pollock
20
EMD Millipore
• Mireille Deschamps
• Juan Costano
• Kenneth Scott
• Jaime Desouza
Asahi Kasei Bioprocess, Inc.
• Todd Nixon
• Hirotomi Naokatsu
Sartorius Stedim Biotech
• Carl Breuning
• Sherri Dolan
Viresolve is a registered trademark of EMD Millipore
Planova is a registered trademark of Asahi Kasei Bioprocess, Inc.
Virosart is a registered trademark of Sartorius Stedim Biotech
21. • The Viresolve® pre-filter + Pro virus filter and Planova® 20N filter
setups demonstrated the lowest flow decay among the molecules
evaluated
21
N/A
22. • The Viresolve® pre-filter + Pro virus filter setup demonstrated the best
overall performance for all molecules evaluated
22
* Vmax could not be calculated due to 0% flux decay
N/A
*