Immunology’s fickle predictability combines with inadequate analytical technology to create a Gordian Knot. This is the modern opportunity for an insightful person or team to step in to solve the problems of immunology associated with particulates in therapy and diagnostics. Organized by IBC Life Sciences (Westborough, MA), the Next Generation Protein Therapeutics Summit was held June 4‒6, 2014, at the Grand Hyatt Hotel in San Francisco, CA. One track focused on Protein Aggregation, Stability, and Solubility. Clearly, this topic is an intense, almost frantic, work in progress.
Immunology of protein therapeutics
Dr. Vibha Jawa of Amgen (Thousand Oaks, CA) lectured on the immunology of protein biotherapeutics. Human immunology is so diverse that nearly every variable, even seemingly minor ones, act to trigger an immune response from someone. If the adverse response is rare, it may be missed in clinical trials. Plus, people have a limited symptom range (rash, fever, mood change, etc.), which complicates connecting the dots between symptom and cause. Animal studies are of limited value since immune systems are generally quite different.
Regulators are stuck, since the number of variables is very high, particularly for aggregation. One question at a lunch roundtable was: What are the limits we must meet? Other than shrugs, the only oral response started with, “That depends….” Later in the day, Prof. Tudor Arvinte (University of Geneva, Switzerland) advised that, in the absence of definitive techniques, protein characterization is derived through a convergence of evidence from many orthogonal methods leading to an unmistakable conclusion.
Arvinte stressed that, “The analysis of protein properties and drug delivery systems requires specific ‘tailor-made’ analytical methods.” Methods should fit the purpose. Some examples: 1) fluorescent stains such as Nile red for detection of nonpolar aggregation, 2) field flow fractionation (FFF) without cross flow, 3) scanners for in situ measurement of vials and storage syringes, and 4) robotics for controlled and reproducible reconstitution. Prof. Arvinte formed Therapeomic, Inc. (Basel, Switzerland) to develop and support these products.
In more detail, protein aggregation or interaction can be caused by interactions of hydrophobic patches. These can be probed with Nile red dye, which fluoresces only in a nonpolar environment. Field flow fractionation without cross flow was useful to show that herceptin should not be reconstituted with 5% dextrose solution since it accelerates aggregation. Prof. Arvinte showed that reconstitution protocols need to be followed exactly. He also developed a robotic system that can easily control shake time and intensity, inversion, repeated aspiration, and dispense cycles, etc., which improves consistency and reduces degradation. Several hundred have been placed.
No measurement technology for particles spans the range of concern in particle size and concentration. The largest, which are visible, should be nondetectable. The remaining subvisible region starts at about 100 μm and extends down to molecules, including molecular clusters, potentially subnanometer. As described in a recent review,1 several techniques provide partial answers, but below about 100 nm, none really provides a quantitative output of population vs size increment. Plus, most work in dilute solutions only. The concentration range of injectables is often above 100 mg dissolved solids/mL. Typical dose is 1 mL. Dilute solution studies using HPLC and FFF have limited relevancy. Absence of aggregate formation in dilute solution does not mean that it will not happen in more concentrated solution. And, lyophilization, even with freeze-drying, involves very concentrated solutions during the last phase of drying.
What do the regulators say?
The regulatory guidance breaks up the problem by particle size range. Dr. Wei Wang of Pfizer (Groton, CT) showed that the FDA’s guidance documents are not consistent across the subvisible particle size range. Many of the guidance documents describe a concern that must be addressed. Specific guidance as to methods and results is noticeably absent. My read is that the FDA is aware of the inadequacies of the best available technology. By posing the problem, the FDA is leaving it to industry to untie the Gordian Knot. After all, industry (pharma and vendors to pharma) employs tens of thousands of scientists. When the knot is untied, I expect that the FDA will revise the guidance documents to reflect advances in technology. This will probably be a stepwise process spanning decades or longer.
Hetero protein aggregates
Hetero protein aggregates typically arise from proteins adsorbing to an active surface. The surface can be glass, which favors more polar interactions, and silicon oil, which leads to assemblies held together by hydrophobic forces. In a lunch discussion roundtable, Prof. Ted Randolph (University of Colorado, Boulder) reminded me of studies on the instability of glass surfaces in vials, particularly on wetting after dehydration. Small pieces of glass spall off the interior surface, which exposes two new surfaces, which are probably polar. The suspended particles may adsorb small amounts of protein via nonspecific adsorption. Depending on size and morphology, these could lead to immune response or mechanical blockage of blood flow. Recall that the adjuvant effect was traced to dirt in poorly cleaned equipment used to produce the first vaccines.
Silicon oil residue is ubiquitous in drug production. One new example is in prefilled syringes for injectables that are difficult to reconstitute or suffer from unstable glass above. In prefilled syringes, silicone oil is the preferred lubricant between the barrel and the plunger. Microscopic examination of the filled product often shows material centered around or connecting small spheres. The spheres are usually attributed to residual silicon oil.
Prof. Arvinte described a laser scanner developed in his laboratory that images the contents of vials and prefilled syringes. If one can measure particulates in the finished product, this may lead to improved stability and shelf life.
The meeting was my first opportunity to see Wyatt Technology’s (Santa Barbara, CA) μDAWN™ MALS detector for ultrahigh-performance liquid chromatography (UHPLC). This detector provides absolute molar mass and size distributions for analysis of protein conformation and conjugate/antibody ratio. The key is a new low-dispersion flow cell and faster electronics.
Measurement of particle size distribution
Since I started to hear concerns about protein aggregates, I wondered if some existing instrument could be repurposed to fill the void. Thus, I was unusually receptive to the lecture by Ms. Denise Nawrocki (Department of Viral Vaccine Research, Merck Research Laboratories, Merck & Co., West Point, PA) on the use of the LUMiSizer® (LUM GmbH, Berlin, Germany) for measurement of particle size distribution in the range of 20 nm to 100 μm. The LUMiSizer is also compatible with concentrated solutions as high as 80% dissolved/suspended solids. Prof. Arvinte privately assured me that it works well for paints, polymers, and other suspensions.
The LUMiSizer starts with a centrifuge. Glass sample tubes or vials are mounted in a horizontal rotor. As the rotor spins, the centrifugal force segregates the sample by density in the sample tube. As the tube spins, a laser measures the optical profile along the major axis of the tube. Differences in optical properties caused by differences in structure are recorded optically as a profile along the major axis of the sample tube.
Ms. Nawrocki compared results from the LUMiSizer with light scattering (LS). In the case studies she presented, LS generally gave values as much as 10 times higher in mass (particle or polymer) than the LUMiSizer, but LS has a reputation for responding preferentially to the largest particles in the detection cell. Minimum concentration is 0.00015 vol%, sample volume is 0.05 to 2.0 mL, and observation time is 1 sec to 99 hr. The rotor can handle 12 tubes or less on each run. Samples are placed in the tube and spun. Samples are recoverable and can be remixed. Despite the differences in size estimation, the LUMiSizer supported drawing the same qualitative conclusions.
Nawrocki reported that the LUMiSizer is particularly useful in studying formulations that often contain excipients to improve stability of a vaccine drug product. The high speed, compared to light scattering, reduced study time by about 25%. A second test involved horizontal rotation of prefilled syringes as part of a shipping test protocol (International Safe Transit Association procedure 3A [ISTA 3A]). Results showed that 15% excipient provided the best stability. Another case study reported a 30% reduction in time/assay in a freeze‒thaw stability study.
While storage can produce particulates, Dr. Francis Kinderman of Amgen showed that the injection process itself can also produce particulates in vivo. Many biotherapeutics are concentrated low-pH cocktails, including stabilizing excipients that improve the solubility of the drug active. Upon injection, the pH abruptly changes, plus the excipients can separate from the drug, leading to precipitation. This was confirmed by fluorescence imaging of injections in rodents.
Dr. Kinderman developed a quick surrogate assay to test for potential precipitation problems. The assay involves injecting an aliquot into a vial of phosphate buffered saline (PBS) and incubating at 37 °C for 24 hr, followed by nephelometry.
Most of the attention on aggregates was on how to avoid them. However, Prof. Arvinte pointed out that some drugs are more effective if they aggregate in vivo. Cases in point are human calcitonin and long-acting insulin analogs. The latter precipitates upon subcutaneous injection releasing insulin over an extended time compared to a fast-acting insulin alternative that does not precipitate in vivo.
The mechanism of protein aggregation was discussed in several lectures with a design goal of developing a workaround to avoid problems and use chemistry to enhance biotherapeutics. For example, Prof. Peter Tessier (Rensselaer Polytechnic Institute, Troy, NY) described modifying the complementary determining regions (CDRs) of antibodies to improve stability by adding charge-inducing amino acid to the CDR. Negative charges are generally more effective than positive. Interestingly, the charges do not seem to affect folding.
Prof. Jennifer Laurence (University of Kansas, Lawrence) used nuclear magnetic resonance (NMR) to locate site-specific changes in protein structure that may precede aggregation. She noted that the free energy of protein folding is only about ‒7 Kcal/mol. Heteronuclear NMR involving 1H and 15N is often useful since it shows large changes with changes in protein conformation. Exchange of 1H and 2H is also useful, since it locates exposed and hence vulnerable amino acids. Both can reveal induced changes in the chemical structure, as does melting point.
PEGylation, where polyethyleneglycol (PEG) is covalently bonded to a protein, can reduce homoaggregation and clearance. Bonding often involves reacting a free cysteine with maleimide-PEG. Dr. David Boisert of PerceptiveBIO LLC (Mountain View, CA) presented a case history of controlling aggregation during refolding and purification. The challenge was to refold a 4-helical bundle protein produced in inclusion bodies during process development. The protein was solubilized in 7 M guanidine and then treated with sodium bisulfite to form three intramolecular disulfide bonds. One free cysteine remained as the site for PEGylation. Overall yield of the synthesis and purification scheme was 5.5%. It was clear that the refolding and PEGylation protocols involved extensive optimization, but details were glossed over, and probably compound specific.
As part of the introduction of Dr. Wei Wang, the session chair credited him with co-authoring Aggregation of Therapeutic Proteins.2 The book provides a thorough and concise explanation of protein dynamics leading to protein aggregation.3 Readers of this review who are interested in learning about protein aggregation should start here. Initial chapters provide a basic description of structure and behavior of proteins and thermodynamics of aggregation. Later chapters focus on aggregation of antibodies, experimental characterization, immunology of aggregates, and more.
Protein Aggregation, Stability, and Solubility was one track of a three-track symposium on Protein Therapeutics held at the Next Generation Protein Therapeutics Summit, which attracted almost 400 scientists. Scott Wallask of IBC deserves special credit for organizing the technical program. IBC’s traveling team worked with the hotel staff to provide the creature comforts, including audiovisual, that are essential for a quality experience.
- Wang, W.; Roberts, C.J. Aggregation of Therapeutic Proteins. Wiley, 2010 ISBN 978-0-470-41196-4.
Robert L. Stevenson, Ph.D., is Editor, American Laboratory/Labcompare; e-mail: firstname.lastname@example.org.