COMMENTARIES
Overlooking Subvisible Particles in Therapeutic Protein
Products: Gaps That May Compromise Product Quality
JOHN F. CARPENTER,1 THEODORE W. RANDOLPH,2 WIM JISKOOT,3 DAAN J.A. CROMMELIN,4
C. RUSSELL MIDDAUGH,5 GERHARD WINTER,6 YING-XIN FAN,7 SUSAN KIRSHNER,7 DANIELA VERTHELYI,7
STEVEN KOZLOWSKI,8 KATHLEEN A. CLOUSE,9 PATRICK G. SWANN,9 AMY ROSENBERG,7 BARRY CHERNEY7
1
Department of Pharmaceutical Sciences, Center for Pharmaceutical Biotechnology, Box 238,
University of Colorado Health Sciences Center, Denver, Colorado 80262
2
Department of Chemical and Biological Engineering, Center for Pharmaceutical Biotechnology, University of Colorado,
Boulder, Colorado 80309
3
Division of Drug Delivery Technology, Leiden/Amsterdam Center for Drug Research (LACDR), Leiden University, Leiden,
The Netherlands
4
Department of Pharmaceutics, Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Utrecht, The Netherlands
5
Department of Pharmaceutical Chemistry, University of Kansas, Lawrence, Kansas 66047
6
Department of Pharmacy, Ludwig Maximilians University, 81377 Munich, Germany
7
Division of Therapeutic Proteins, Center for Drug Evaluation and Research, US Food and Drug Administration, Rockville,
Maryland 20857
8
Office of Biotechnology Products, Center for Drug Evaluation and Research, US Food and Drug Administration,
Rockville, Maryland 20857
9
Division of Monoclonal Antibodies, Center for Drug Evaluation and Research, US Food and Drug Administration,
Rockville, Maryland 20857
Received 27 May 2008; revised 2 July 2008; accepted 10 July 2008
Published online 14 August 2008 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21530
INTRODUCTION these treatments are used chronically to slow
disease progression, reduce morbidity and/or to
Therapeutic protein products provide unique and replace essential proteins that are not produced
effective treatments for numerous human dis- endogenously in patients. Therefore, any factor
eases and medical conditions. In many cases, that reduces or eliminates the effectiveness of the
treatment can lead to patient suffering and even
death. One means by which efficacy of therapeutic
The information here reflects the current thinking and
proteins can be compromised is by an immune
scientific judgment of the authors. However, this is not a policy
document and should not be used in lieu of regulations, pub- response, resulting in antibody-mediated neutra-
lished guidance, or direct discussions with regulators.
lization of the protein s activity or alterations
Correspondence to: John F. Carpenter (Telephone: 303-315-
in bioavailability.1,2 For example, in the case of
6075; Fax: [phone]; E-mail: [e-mail])
treatment of hemophilia A, neutralizing anti-
Journal of Pharmaceutical Sciences, Vol. 98, 1201 1205 (2009)
bodies to Factor VIII can cause life-threatening
ß 2008 Wiley-Liss, Inc. and the American Pharmacists Association
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 4, APRIL 2009 1201
1202 CARPENTER ET AL.
bleeding episodes, resulting in significant mor- Furthermore, protein particles (visible and sub-
bidity and necessitating treatment with a pro- visible) can be generated from protein alone or
longed course of a tolerance-inducing therapy to from heterogeneous nucleation on foreign micro-
reverse immunity.3,4 In other cases, drug-induced and nanoparticles that are shed, for example, from
antibodies to a therapeutic version of an endo- filling pumps or product container/closures.6 8
genous protein can cross-react with and neutra- The levels and sizes of protein particles present
lize the patient s endogenous protein. If the in a given product can be changed by many factors
endogenous protein serves a nonredundant bio- relevant to commercial production of therapeutic
logical function, such an immune response can proteins. Such factors include a change in the type
have devastating results. For example, pure red of filling pump during scale-up to commercial
cell aplasia can result from neutralizing anti- manufacturing, changes in formulation or con-
bodies to epoetin alpha.1,2 tainer/closure, and even unintentional changes in
It is well established that protein aggregates in the manufacturing process such as alterations in
therapeutic protein products can enhance immu- filling pump mechanical parameters or other
nogenicity,2 and such an effect is therefore an unforeseen factors.8,9 Thus, unless appropriate
important risk factor to consider when assessing quality controls are in place for subvisible
product quality. The purpose of this commentary particles, a product that was safe and effective
is to accomplish the following: in clinical trials may unexpectedly cause adverse
events in patients after commercialization.
(i) provide brief summaries on the factors
affecting protein aggregation and the key
aspects of protein aggregates that are asso- EFFECTS OF AGGREGATE CHARACTERISTICS
ciated with immunogenicity; ON IMMUNOGENICITY
(ii) emphasize the current scientific gaps in
understanding and analytical limitations From work on fundamental aspects of immunol-
for quantitation of species of large protein ogy and vaccine development, it is known that
aggregates that are referred to as subvisi- large protein assemblies with repetitive arrays of
ble particles, with specific consideration of antigens, in which the protein molecules have
those particles 0.1 10 mm in size; native conformation, are usually the most potent
(iii) offer a rationale for why these gaps may at inducing immune responses.2,10,11 Further-
compromise the safety and/or efficacy of a more, efforts to develop more effective vaccines
product; have shown that adsorbing antigenic proteins to
(iv) provide scientifically sound, risked based nano- or microparticles comprised of other mate-
recommendations/conclusions for assess- rials (e.g., colloidal aluminum salts or polystyr-
ment and control of such aggregate species. ene) can greatly increase immunogenicity.12,13
Applying these lessons to therapeutic protein
products, it has been argued that large aggregates
containing protein molecules with native-like
CAUSES OF PROTEIN AGGREGATION conformation pose the greatest risk of causing
adverse immune responses in patients.2 Thus, for
Proteins usually aggregate from partially example, particles of therapeutic proteins formed
unfolded molecules, which can be part of the by adsorption of protein molecules onto foreign
native state ensemble of molecules.5 Even though micro- and nanoparticles might be particularly
product formulations are developed to maximize prone to cause immunogenicity. These particles
and maintain the fraction of the protein molecules contain numerous protein molecules, and in the
present in the native state, significant amounts of two examples published to date, the adsorbed
aggregates can form, especially over pharmaceu- protein molecules were shown to retain their
tically relevant time scales and under stress native conformations.6,8
conditions. For example, exposure to interfaces Unfortunately, lacking are published studies
(e.g., air liquid and solid liquid), light, tempera- that comprehensively investigate the range of
ture fluctuations or minor impurities can induce parameters that could influence immunogenicity
aggregation. Such exposure can occur during of aggregates. Because each protein may differ in
processing steps, as well as in the final product aggregate formation and consequences, factors
container during storage, shipment and handling. that need to be investigated include but are not
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 4, APRIL 2009 DOI 10.1002/jps
OVERLOOKING SUBVISIBLE PARTICLES IN THERAPEUTIC PROTEIN PRODUCTS 1203
limited to type, amount, and size of aggregates, as Historically, analysis of subvisible (and visible)
well as protein conformation in aggregates, on a particles has been required for final product
case by case basis. Of course, other factors, release testing to mitigate the risk associated
particularly pertaining to patient status and with the presence of extraneous particles in
treatment protocol, are also critical in determin- injectable solutions, particularly the risk of blood
ing the propensity to generate immune responses. vessel occlusion for intravenously administered
These include immune competence of the solutions of small molecule parenteral drug
patients, route of administration, and dosing products. Consequently, USP requirements for
frequency and duration. Given the consequences the light obscuration test <788>, the standard
of aggregate-induced immunogenicity in patients, test for subvisible particulate analysis, specifies
it is important to understand these issues and to that particulates >10 mm in size are controlled at
reduce the risk to product quality for every or below 6000 particles/container and particles
therapeutic protein product. >25 mm are limited to at or below 600 particles/
Because the exact characteristics and levels of container Although ICH quality guidance Q6B
protein aggregates that lead to an enhanced states that the requirements set forth by phar-
immune response are unclear and may differ macopoeias pertaining to analytical procedures
among proteins, it is not possible to predict, and acceptance criteria for particulate matter are
a priori, the in vivo effects of different sizes, types applicable to biotechnological products,14 the
or quantities of aggregates for therapeutic protein risks associated with the administration of large
products. In such situations, careful analysis of aggregated protein particles were never consid-
the relationship between clinical performance and ered in establishment of the USP light obscuration
the presence of protein aggregates in relevant test <788>. However, ICH Q6B clearly states that
clinical trial material may help in the design of specifications   should focus on those molecular
suitable control strategies that ensure product characteristics found to be useful in ensuring
quality. However, the validity and utility such the safety and efficacy of the product  14 and is
correlations are only optimized when the full potentially applicable to the control of large
spectrum of protein aggregate species are thor- protein aggregates.
oughly characterized by multiple and orthogonal In USP 30 monograph <788>, there is an
techniques. exclusion of injections intended solely for intra-
muscular or subcutaneous administration. This
exclusion may be appropriate for the risk of blood
vessel occlusion but would not be appropriate for
CRITICAL GAPS IN THE ANALYSIS AND the risk of immunogenicity. Since subcutaneous
CONTROL OF SUBVISIBLE PARTICLES and intramuscular routes are often more immuno-
genic than intravenous administration, it is
Protein aggregates that can be quantified based appropriate to consider particulates for all par-
on the mass percentage for each aggregate size are enterally administered proteins. The revised
usually classified as soluble and insoluble. There harmonized version of <788> published in USP
is mass balance between the amount of protein in 31 does not contain this exclusion and thus
the aggregates and the loss of the monomeric appears to cover all parenteral routes. However,
protein. However, subvisible particles usually do subvisible particles less then 10 mm are still not
not constitute a sufficient mass fraction of the evaluated in the USP test.
protein population to be quantified based on mass Thus, even though large protein aggregates
of protein in the particles or by loss of monomeric that are classified as subvisible particles are
protein. Typically, these particles are quantified potentially the most immunogenic class of protein
by counting the number of particles in given size aggregates, subvisible particles smaller than
ranges. Subvisible particles are usually defined 10 mm are not currently monitored and recom-
as particles that are too large for analysis mendations for such testing for therapeutic
by size exclusion chromatography (SEC) (e.g., proteins products are lacking. Furthermore,
>0.1 mm), but too small to be visible to the unlike the concern regarding extraneous particles
unaided eye (e.g., <100 mm). Subvisible protein in small molecule parenteral products, protein
particles are thus relatively large assemblies (e.g., particles can accumulate over time during storage
0.1 10 mm) that contain thousands to millions of of the final presentation. Subvisible protein
protein molecules. particles in the 0.1 10 mm range, as well as
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 4, APRIL 2009
1204 CARPENTER ET AL.
protein particles >10 mm, have the potential to is microflow digital imaging, which can be used for
impact the safety and efficacy of the product over particles as small as 0.75 mm. For each of these
its shelf life, a criteria described in ICH Q6B. As types of instruments, there are important sample
with any product attribute, the level of control handling issues including: (a) the potential
necessary should reflect the potential risk to requirement for a sample volume exceeding the
product quality and could involve monitoring at unit dose volume (e.g., 1 mL) for a given protein
release and on stability. Risk-mitigating factors product; (b) interference from air bubbles; (c)
are important in assessing the impact of a product potential need to dilute a sample; and (d) dif-
attribute. For aggregates, such a factor could be ficulties in handling high viscosity samples.
the reversibility of the aggregate under the Furthermore, because protein particles may be
intended route of administration. translucent and loosely packed, compendial light
The need for manufacturers of pharmaceutical obscuration techniques and the other methods
protein products to evaluate and control this risk may not work as well to quantify protein particles
to product quality is underscored by published as they do for particles from extraneous materials.
case studies with therapeutic proteins in which More research is needed to rigorously assess the
subvisible particles were measured.8,15,16 For capabilities of current instruments to quantify
example, during filling pump operations, it was protein particles as small as 0.1 mm and to develop
found that an IgG formed hundreds of thousand of appropriate low volume sampling methods. Micro-
particles per mL in the 1.5 3 mm size range, but scopic techniques such as atomic force microscopy
less than 1000 particles per mL in the 8 15 mm and electron microscopy may be valuable, but
size range.8 In a formulation study of a ther- these are low throughput methods and also would
apeutic cytokine formulated with human serum need extensive method validation. Finally, con-
albumin, more than 90% of subvisible particles sidering these technical issues, using multiple
were in the 1 2 mm size range, 7 9% were 2 and orthogonal methods may currently be the
10 mm and less 0.01% were larger than 10 mm.15 most prudent means for evaluating subvisible
Similarly, subvisible particle counts in a recombi- particulates.
nant hemoglobin product documented that parti-
cles smaller than 10 mm were approximately two
orders of magnitude greater in number than CONCLUSIONS
particles larger than 10 mm.16 As demonstrated by
these examples, protein products that contain (1) Subvisible protein particles have the poten-
extremely large numbers of particles smaller tial to negatively impact clinical perfor-
than 10 mm might meet the current USP mance to a similar or greater degree than
requirement for subvisible particles. If only other degradation products, such as soluble
particles >10 mm were quantified in a given aggregates and chemically modified species
product, there could be gaps in understanding of that are evaluated and quantified as part of
important degradation products and in product product characterization and quality assur-
quality assessment. ance programs.
Also, there are important technical issues (2) Current USP particulate testing is not
related to counting protein particles, especially designed to control the potential risk of
those smaller than 1 mm, which is close to the large protein aggregates to impact protein
lower size limit of detection of particle counting immunogenicity. Analytical methods that
instruments operating by light obscuration or can assess particulate characteristics
electrical current sensing zones. Hence, it is not (including composition, amount and rever-
clear if the results obtained for the smaller size sibility of the protein aggregate) are critical
range of particles (e.g., 1 mm) are of sufficient for developing scientifically sound ap-
accuracy and precision to allow for method proaches for evaluating and mitigating risk
qualification and, more stringently, validation of to product quality caused by large protein
the method for use in pharmaceutical quality aggregates.
assurance. Particle counters that operate by laser (3) Pharmaceutical and academic researchers
light scattering can quantify particles as small as and instrument manufacturers should work
0.1 mm, but there have not yet been published together to help define the quantitative
studies documenting the utility of these systems capabilities of current particle counting
for protein particles. Another alternative method instruments for particles as small as
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 4, APRIL 2009 DOI 10.1002/jps
OVERLOOKING SUBVISIBLE PARTICLES IN THERAPEUTIC PROTEIN PRODUCTS 1205
0.1 mm and develop new instruments as manufacture of protein pharmaceuticals. New
York: Kluwer Academic/Plenum Press. pp 47 127.
needed.
8. Tyagli AK, Randolph TW, Dong A, Maloney KM,
(4) The impact of protein aggregates on im-
Hitscherich C, Jr., Carpenter JF. 2008. IgG particle
munogenicity needs to be elucidated and
formation during filling pump operation: A case
should include studies of the role of protein
study of heterogeneous nucleation on stainless
class, amount of aggregate, size of aggre-
steel nanoparticles. J Pharm Sci, DOI: 10.1002/
gates, and protein conformation in aggre-
jps.21399.
gates. These investigations should be
9. Cromwell ME, Hilario E, Jacobson F. 2006. Protein
published in peer reviewed journals.
aggregation and bioprocessing. AAPS J 8:E572
E579.
10. Hermeling S, Aranha L, Damen JM, Slijper M,
Schellekens H, Crommelin DJ, Jiskoot W. 2005.
Structural characterization and immunogenicity
REFERENCES
in wild-type and immune tolerant mice of degraded
recombinant human interferon alpha2b. Pharm
1. Kessler M, Goldsmith D, Schellekens H. 2006. Res 22:1997 2006.
Immunogenicity of biopharmaceuticals. Nephrol 11. Spohn G, Bachmann MF. 2008. Exploiting viral
Dial Transplant 21:9 12. properties for the rational design of modern vac-
2. Rosenberg AS. 2006. Effects of protein aggregates: cines. Expert Rev Vaccines 7:43 54.
An immunologic perspective. AAPS J 8:E501 E507. 12. Lebron JA, Wolf JJ, Kaplanski CV, Ledwith BJ.
3. Hooks WK. 2006. Urgent inhibitor issues: Targets 2007. Nonclinical safety assessment of vaccines and
for expanded research. Haemophilia 12:107 113. the evaluation of novel adjuvants and delivery sys-
4. Reipert BM, van den Helden PM, Schwartz H-P, tems. In: Singh M, editor. Vaccine adjuvants and
Hausl C. 2007. Mechanisms of action of immune delivery systems. New York: Wiley. pp 403 420.
tolerance induction against factor VIII in patients 13. Xiang SD, Scholzen A, Miningo G, David C, Apos-
with cogenital haemophilia A and factor VIII inhi- tolopoulos V, Mottram PL, Plebanski M. 2006.
bitors. Br J Haemophilia 136:12 25. Pathogen recognition and development of particu-
5. Chi EY, Krishnan S, Randolph TW, Carpenter JF. late vaccines: Does size matter? Methods 40:1 9.
2003. Physical stability of proteins in aqueous solu- 14. ICH Q6B Test Procedures and Acceptance Criteria
tion: Mechanism and driving forces in nonnative for Biotechnological/Biological Products 8/18/
protein aggregation. Pharm Res 20:1325 1336. 1999.
6. Chi EY, Weickmann J, Carpenter JF, Manning MC, 15. Hawe A, Friess W. 2007. Stabilization of a hydro-
Randolph TW. 2005. Heterogeneous nucleation- phobic recombinant cytokine by human serum
controlled particulate formation of recombinant albumin. J Pharm Sci 96:2987 2999.
human platelet-activating factor acetylhydrolase 16. Kerwin BA, Akers MJ, Apostol I, Moore-Einsel C,
in pharmaceutical formulation. J Pharm Sci 94: Etter JE, Hess E, Lippincott J, Levine J, Mathews
256 274. AJ, Revilla-Sharp P, Schubert R, Looker DL. 1999.
7. Akers MJ, Vasudevan V, Stickelmyer M. 2002. Acute and long-term stability studies of deoxy
Formulation development of protein dosage forms. hemoglobin and characterization of ascorbate-
In: Nail SL, Akers MJ, editors. Development and induced modifications. J Pharm Sci 88:79 88.
DOI 10.1002/jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 4, APRIL 2009