Sunteți pe pagina 1din 10

Scaling Down

of Biopharmaceutical
Unit Operations –
Part 2: Chromatography
and Filtration
Anurag Rathore,
Raj Krishnan,
Stephanie Tozer,
Dave Smiley,
Steve Rausch, and
Jim Seely
2005;18(4):58–60,62,64

Constraints associated with equipment can make


scale-down a challenging exercise.

C
reation and qualification
of scale-down models is
Chromatography Scale-down
essential for performing In theory, chromatography is one of
several critical activities the more straightforward unit opera-
that support process validation and tions to scale down. Factors at the
commercial manufacturing. Part 1 large scale such as column-header
covered fermentation. In this seg- design and packing can be difficult
ment, we present some guidelines to control and may result in reduced
and examples for scale-down of accuracy of the scale-down model.
common downstream unit opera- This section presents a review of gen-
tions used in biotech processes — eral guidelines for scaling down.
chromatography and filtration. Use representative feed streams,

100 | BioPharm International


preferably from full-scale manufactur- ies. It offers precise, accurate, and
ing. First perform a complete analy- reproducible control of flow rates
sis of product concentration, quality and gradients. However, you still
attributes, and other feed stream have to check and calibrate the accu-
characteristics (such as pH and con- racy and precision of in-line detec-
ductivity) to ensure a good match. If tors (pH, conductivity, temperature)
possible, use feed streams from differ- by using off-line probes.
ent lots at small scale to learn about The type of column is generally
the effect of feed stream quality on not a major issue; however, when-
the performance. In addition, acquire ever possible use the same bed sup-
some knowledge of the storage and port material. We prefer to not use
stability of the feed stream material. column diameters of less than 1.0
Storage-related degradation of the cm since the ratio of system dead
inventory could affect the outcome volume to column volume can result
of the small-scale runs. in pool concentrations significantly
Use a representative chromatogra- lower than those seen at large scale.
phy resin. It would be ideal to have The packing procedures and solu-
resin from the lot being used at full tions should be similar to those used
scale. If you can’t obtain GMP-released at large scale.
resin, then make a match based on the Controlling temperature at small
vendor’s certificate of analysis. scale to mimic that of cold rooms
Buffers and other solutions should or cold-boxes can be a challenge.
be made up and released by man- The use of water baths with col-
ufacturing. If it is not possible to umn jacketing can provide precise
obtain these, mix solutions with temperature control, but the tub-
GMP-released raw materials, or mate- ing leading up to the column can
rial of an equivalent grade from an frequently act as a mini heat
approved vendor. Follow exactly the exchanger, making it difficult to
solution recipes from manufacturing. control the actual inlet temperature.
The pH, temperature, and conductiv- Using heat exchangers on the inlet
ity probes that check buffer properties tubing and column jacketing may
must be comparable with those used be necessary for accurate, consistent,
at the large scale. Measure pH and and precise temperature control.
conductivity at the same temperature
spelled out in the recipe. The pH and Operational Equivalence
conductivity standards should also be The general approach towards quali-
comparable (preferably identical) to fication of a scale-down model is
those used in manufacturing. to run all operational parameters
(inputs) at the center of the operat-
Hardware ing range used at large scale (clini-
Our experience has been that an cal or commercial manufacturing).
automated small-scale chromatog- The performance parameters (out-
raphy system, such as the ÄKTA puts) such as step yields, pool purity,
Explorer, excels at small-scale stud- impurity clearance, elution volume

Elements of Biopharmaceutical Production Series | 101


and retention time should be similar should be shown to be representa-
across scales. Some of the significant tive of commercial scale manu-
operational parameters include: facturing. If this is not possible
1

• Column loading (g protein/L resin) and the bed height is different at


should be the same at both scales. small scale, the linear flow velocity
In size-exclusion chromatography, should be adjusted appropriately to
the loading should be based on a keep the residence time constant. 2

volume percentage of the column • Pooling criteria should be the same


bed volume. as in manufacturing. They can be
• Gradient lengths (in number of col- online (conductivity or absorbance)
umn volumes) and slopes (increase or offline (gels or HPLC). In our
in percent limit buffer per unit experience, if the pooling criteria
time, or similar measurement) are UV-based, it is a good practice to
should be the same for all steps of a ensure that the UV detectors of the
chromatography procedure such as small-scale systems correlate well
equilibration, wash, and elution. with those in the large-scale system.
• Temperature, pH, and conductiv- UV detectors may be inaccurate or
ity should be identical for all buf- non-linear over the optical density
fers and feed streams. Differences (O.D.) ranges required. If the absor-
in temperature across scales can bance range is non-linear at 280
lead to significant changes in con- nm, it may be prudent to measure
ductivity of some buffers that may the on-line O.D. at 300 nm or some
impact step performance. other wavelength that has a lower
• Bed height of the chromatography extinction coefficient.
column and the linear flow velocity
are generally kept constant across Model Qualification
scales. The ICH guideline on viral Pre-determine the acceptance criteria
safety states that the validity of the prior to the scale-down runs. These
scaling down should be demon- can be based on a statistical compari-
strated, and that column bed height son of the small-scale data against the

Table 1. Comparison of Scale-down Model to Large-scale Cation-exchange Column.3


Input material roughly 98 percent antibody, 3 ppm DNA, 3,000 ppm CHOP and <7.8
ppm Protein A. (Reference 3 shows intermediate cycles.)

SCALE Cycle # Yield, % CHOP,* DNA, ppm Protein A,


ppm ppm

Small-scale 1 88.7 75 < 0.006 < 7.8


51 88.3 34 0.01 < 7.8

Manufacturing 1 84.6 116 < 0.07 < 7.8


Scale 50 81.1 72 < 0.03 < 7.8

*Chinese hamster ovary proteins (host-cell proteins)

102 | BioPharm International


full-scale data using a t-test, tolerance differences such as header effects
intervals, or other statistical tests. and pressure vs. flow characteristics
Alternatively, it may be demonstrated may result in anomalies with regard
that the small-scale data are within to these parameters. You must be
the full-scale historical range. The able to explain these scale-depen-
approach taken may depend on the dent differences.
inherent variability of the assays used If significant differences exist
to determine quality attributes or the between the two scales, make a care-
amount of large-scale data available. ful determination of the impact of
Sometimes there may be no large- these differences and if necessary,
scale data available at this stage of redesign the small-scale model. In
the project and the qualification of any case, having a qualified scale-
the small-scale model may have to be down model prior to performing
done retrospectively after data from small-scale cycling studies is neces-
large scale are available. It is a good sary, and using a scale-down model
practice to perform several small-scale that is flawed will lead to unreliable
runs (at least three), to better under- lifespan study results and a waste of
stand the inherent variability of the time and resources.
process, particularly when there are Table 1 presents an example of
limited amounts of large-scale data a successful scale-down of a cat-
available for comparison. ion-exchange chromatography step
The model should achieve the and its use for measuring column
same level of separation as at large lifespan. This column is perform-
3

scale. Aim for a similar elution ing purification in production of a


profile. Deviations that cannot be monoclonal antibody. The input val-
avoided should be analyzed care- ues are roughly 3,000 ppm Chinese
fully with regard to influence on the Hamster Ovary Proteins (CHOP), 3
results. Step yields and the quality ppm DNA, and <7.8 ppm Protein
attributes of the product (host cell A. At both scales the step yields are
proteins, host cell contaminants, comparable and the key functions of
product related variants, and aggre- the column step, which are reducing
gates) should be comparable to full- CHOP and DNA and clearing Protein
scale. Height-equivalent theoretical A, are equivalent at both scales.
plates (HETP) and asymmetry values Figure 1 presents another compari-
should also be within the accept- son of the chromatographic profiles
able range of that for the large-scale at small and large scale. The pro-
columns (although not necessarily file (absorbance) and the resolution
identical). between the product and the dif-
Take data and evaluate the ferent impurities are similar across
importance of other parameters scales. Impurity concentrations were
that indicate column performance, higher in the feed material for the
including pool volume, peak reten- small-scale run due to under-perfor-
tion time, and peak shape. System mance of the preceding step.
dead volume and unavoidable scale Figure 2 illustrates an example of

Elements of Biopharmaceutical Production Series | 103


Main Peak - LS (percent) Main Peak-SS (percent)
Absorbance - LS (mAU) Absorbance-SS (mAU)

100 3
90
80 2.5
Purity, Percent

70 2

Absorbance
60
50 1.5
40
30 1
20 0.5
10
0 0
0 10 20 30 40
Sequential Sample Number

Absorbance - LS (mAU) Absorbance - SS (mAU)


Impurity 1 - LS (percent) Impurity 2 - LS (percent)
Impurity 3 - LS (percent) Impurity 1 - SS (percent)
Impurity 2 - SS (percent) Impurity 3 - SS (percent)
3 14

2.5 12
Impurity, Percent

10
Absorbance

2
8
1.5
6
1
4
0.5 2

0 0
0 10 20 30 40
Sequential Sample Number

Figure 1. Chromatographic Profiles of Small- and Large-scale


Columns. Raw material is a proprietary aqueous solution that
has undergone one step of preliminary purification. The small
column (green data) is 1.6 cm dia by 12 cm. The large column
(black data) is 63 cm dia by 12 cm.

104 | BioPharm International


a successful scale-up from lab-scale lobe pump is used at large-scale, do
to pilot-scale facility as the column not use a high-shear gear pump at
performance improved both in small scale.
terms of step yield and pool purity.4 Select membranes from a lot that
However, if one were to evaluate the is representative of full-scale produc-
performance in the lab with respect to tion. For new membranes, it is a good
qualifying the scale-down model, dif- practice to utilize normalized water
ferences in step yield and pool purity permeability (NWP) as a benchmark
would necessitate modification of the for their integrity. Before running the
scale-down model. system, run a baseline of the permeate
at 280 nm and a pressure-hold test to
Filtration Scale-down ensure the system is not leaking.
The unit operation of filtration is per- When making tests, run all opera-
haps the most common unit opera- tional parameters at the center point
tion that is used in bioprocesses. It is of the operating range used for the
used as depth filtration for separation large-scale filtration. These include:
of particulates (cell debris) from the retentate cross-flow rate, transmem-
process stream; as normal flow filtra- brane pressure (TMP), both inlet
tion (NFF) for separation of impuri- and outlet pressures, temperature,
ties (viruses, host cell proteins, DNA) number of diafiltration volumes,
from the process stream; as ultrafiltra- and concentration factor. Crossflow
tion (UF) for concentration of the feed and transmembrane pressure are the
stream; and as diafiltration (DF) for two main drivers for removal of low
buffer exchange. Here is a review of molecular weight solutes and concen-
general guidelines to consider while tration of the protein. Also make sure
scaling down a filtration step. Many of the membrane loading (g product/m2
these guidelines apply to all formats, membrane) is the same across scales.
but some may be specific for a certain Adequate stirring of the retentate res-
format. It is up to the reader’s discre- ervoir is important to ensure the dia-
tion to apply these appropriately. filtration is as complete as possible.
As with the chromatography steps,
use feedstock from large-scale GMP Model Qualification
manufacturing runs (or equivalent). Pre-determine the acceptance criteria
Buffers should either be made in prior to the scale-down runs. These
GMP manufacturing or according to can be based on a statistical compari-
the manufacturing recipes. son of the small-scale data against the
The retentate path-length of the full-scale data. Be cautious — as the
equipment should be the same at system dead volume is almost always
both scales. The Millipore Pellicon larger at the small scale and will result
XL format scales from 50 cm 2 to in lower pool concentrations or lower
2.5 m2 and the Pall Centramate scales step yields. Deviations such as these
from 0.1 to 50 ft2. The type of pump should be kept under consideration
used should be of the same shear when evaluating results from the
characteristics. If a low-shear rotary- scale-down model and also when set-

Elements of Biopharmaceutical Production Series | 105


LAB SCALE
Dimensions: 1.6 cm x 5 cm (10 mL)
Pool purity: 94% (by AE-HPLC)
Yield: 51%

100

Prepeak Impurity
80
AE-HPLC Purity %

Product
Postpeak Impurity
60

40

20

0
2 6 10 14 18 22 26
Column Fraction

PILOT SCALE
Dimensions: 20 cm x 20 cm (6.3 L)
Pool purity: 99% (by AE-HPLC)
Yield: 61%

100
90
80 Product
AE-HPLC Purity %

70 Impurity
Impurity*
60
50
40
30
20
10
0
1 6 11 16 21

Column Fraction

Figure 2. Example of a Successful Scale-up but an Unsuccessful


Scale-down for a Chromatography Step.4 Feedstock is an aqueous
solution. AE-HPLC = Anion exchange high-pressure liquid chromatography

106 | BioPharm International


Figure 3. Comparison of Performance of a Depth Filtration
Step at Small-scale and Large-scale.5

Elements of Biopharmaceutical Production Series | 107


Table 2. Comparison of Small-scale Disc Format with Large-scale Cartridge Format.6
Aqueous solution feed for a NFF step for viral clearance

Discs Cartridges
Lot 1 Lot 2 Lot 3 Lot 1 Lot 2
Average Process Flux 75.3 70.6 74.6 75.1 74.9
(L/(m2-hr)
Protein Recovery, percent 98 97 98 100 100

øX174 LRV 4.0 4.2 4.5 4.0 4.0

ting the acceptance criteria. requirements can be challenging.


TMP vs. flux curves are useful to In general, downstream unit opera-
compare the decay in flux over pro- tions are easier to scale down than
cessing time. In cases where the UF upstream ones. Exceptions to this
or DF step affects product quality, are protein-refolding unit operations
a measurement of these key quality that may require more extensive
attributes should be included. modeling efforts.7 These models are
Diafiltration efficiency is impor- useful for performing experimental
tant for DF applications and should studies in an efficient and economi-
include measurements of diafiltration cal fashion.
volumes vs. conductivity and pH. A However, there are consider-
measurement of excipient concen- ations unique to each unit opera-
trations may be appropriate for final tion that apply during scale-down.
formulation steps. Con-straints associated with equip-
Figure 3 illustrates a scale-down ment or methodology may prevent
of a depth filtration step. The 23- a “true” scale-down of a unit opera-
cm 2 model closely mimics the tion. Dead volumes and hold times
performance at 1.8 m2 in the operat- are typically greater at manufacturing
ing range with respect to the pres- scale. Path lengths of detectors typi-
sure drop and the filtrate turbidity cally vary between instruments at
vs. throughput.5 Table 2 shows data the lab and the manufacturing plant.
supporting the use of filter discs for Wetted materials (gaskets, valves,
scale-down modeling of large filter and pumps) are not likely to be the
cartridges using the øX174 bacterio- same at two scales. Differences in cas-
phage model.6 Flux, yield, and viral sette design between two scales may
clearance are shown to be comparable impact the outcome of lifetime stud-
across the scales. ies of UF and DF membranes.
Scale-down studies certainly
Summary increase the chances for a successful
Creation of scale-down mod- validation campaign, but they only
els that can meet the qualification provide “guidance.” The actual “con-

108 | BioPharm International


firmation” must be made at large
3. O’Leary RM, Feuerheim D, Peers D,
scale. The only exception to this is Xu Y, Blank GS. Determining the use-
the viral-clearance validation studies ful lifetime of chromatography resins,
that are performed at small scale. ◆ Biopharm Int’l. 2001 Sept.; 14 (9):10-
18.4.

References 4. Rathore AS, Velayudhan A. Scale-up


guidelines for preparative chromatog-
1. ICH, Quality of biotechnological prod- raphy, Biopharm Int’l. 2003 Jan.; 16
ucts: Viral safety evaluation of biotech- (1):34-42.
nology products derived from cell lines
of human or animal origin. International 5. Rathore AS, Wang A, Menon M, Riske
Conference on Harmonization Step 4 F, Campbell J, Goodrich E, Martin J.
Endorsement 1997 March. Available at: Optimization, scale-up and validation
www.fda.gov/cder/guideance/Q5A-fnl.pdf. issues in filtration of biopharmaceu-
ticals – Part I. Biopharm Int’l. 2004
2. Yamamoto S, Nomura M, Sano Y, August; 17 (8):50-58.
Resolution of proteins in linear gradi-
ent elution ion exchange and hydro- 6. Lutz H, Siwak M, Carter J. Viral
phobic interaction chromatography. clearance performance and regula-
J. Chromatogr. 1987:409:101-110. tory requirements, Poster, Biophex/
O’Leary RM, Feuerheim D, Peers D, Interphex, 2002 October 23-24; Santa
Xu Y, Blank GS. Determining the use- Clara, CA.
ful lifetime of chromatography resins, 7. Rathore AS, Sharma A, Zhang RJ, Chilin
Biopharm Int. 2001 Sept.; 14 (9):10-
D. Scale-down modeling and character-
18.4.Rathore AS, Velayudhan A. Scale-
ization of a protein refolding step, Oral
up guidelines for preparative chroma-
presentation, 227th American Chemical
tography, Biopharm Int’l. 2003 Jan.; 16
Society National Meeting, 2004 March
(1):34-42.
28-April 1; Anaheim, CA.

Originally published in the April 2005 issue of BioPharm International.

Elements of Biopharmaceutical Production Series | 109

S-ar putea să vă placă și