Silicone Microdroplets in Protein Formulations—Detection and Enumeration

[This article was originally published on PharmTech.com in January 2009. It was later published in the April 2009 issue of Pharmaceutical Technology; Vol. 33, Issue 4, pp. 74-79.]

Silicone oil, because of its inert biologic properties, is widely used in the pharmaceutical industry as a lubricant for rubber stoppers, plungers, and prefilled syringes (1). Silicone oil is applied in these syringes to facilitate easy movement of the plungers within the barrels. One consequence of using silicone oil is the potential for microdroplets to shed from the coating into the drug formulation. Biopharmaceutical products are sensitive to silicone oil because it can interact with protein-based drugs or active ingredients producing particles (2–7). The presence of silicone oil droplets in parenteral formulations is also suspected of causing protein aggregation (4, 7). The potential impact of silicone oil contaminants on drug products is possible in some cases for certain sensitive therapeutic proteins and antibodies, especially with active biopharmaceutical drug products that may contain mere femptograms of active ingredient. As a result, quantification of silicone oil droplet populations in chemical and pharmaceutical products may be important. When these products also contain populations of particles composed of other materials such as protein aggregates, this is a challenging assignment.

United States Pharmacopeia General Chapter <788> and the harmonized versions in the European and Japanese Pharmacopoeias (8–10) set limits and cite enumeration methods for subvisible, foreign particulate matter in parenteral products. USP <788> does not consider intrinsic sources of particulate matter such as protein aggregates, silicone oil droplets, air bubbles, micelles, or precipitates. However, the test methods cited in the section, which include automatic light obscuration and filter membrane microscopic analysis, might also in principle be applied to mixed silicone droplet and aggregate populations. However, both particle types are highly transparent and not easily measured by either of these techniques. In addition, automated light obscuration provides no capability, and membrane microscopy limited capability, to differentiate between these two highly transparent particle types.

Micro-Flow Imaging

Micro-Flow Imaging (MFI, Brightwell Technologies Inc., Ottawa, ON, Canada) uses digital imaging of a flowing sample stream for analyzing particles in suspension. Brightfield images of individual particles are captured as a sample stream passes through a flow-cell centered in the field of view of a custom magnification system having a well-characterized and extended depth of field. Images are analyzed by the system software to extract each particle. The software compiles a database containing count, size, and concentration as well as morphological parameters like intensity, circularity, maximum Feret's diameter and aspect ratio. This database is analyzed using the system software to produce particle distributions and isolate particle subpopulations using histograms and scatter plots of particle statistics (11). The instrument has a particle size range of 2 μm to 300 μm for low magnification (400 μm flow cell) and 0.75 μm to 75 μm for high magnification (100 μm flow cell) systems. Particle subpopulations in parenteral samples are isolated, quantified, and characterized by applying custom-designed morphological filters using parameters such as aspect ratio, circularity, and intensity. MFI complements <788> techniques by assisting in classifying the different extrinsic and intrinsic particle types present in injectable solutions.

Experiment description

Materials. Monoclonal IgG1 antibody (>95% purity, Cat # I5154) and silicone oil (Dow Corning Corporation "200," 1000 cSt) was obtained from Sigma-Aldrich (Oakville, ON, Canada). Dulbecco's phosphate buffered saline (without calcium chloride and magnesium chloride, pH 7.2) was obtained from Gibco (Carlsbad, CA). Filtered (0.22 μm Durapore membrane filters, Millipore, MA) de-ionized water was used to make all solutions.

Methods. Silicone oil emulsion. A silicone oil suspension of 1.5% (w/w) in phosphate buffered saline (PBS) buffer was prepared by gravimetric dilution in a 50 mL polypropylene centrifuge tube. The solution was sonicated for 30 minutes in a Bransonic 1200 (Branson, CT, USA) ultrasonicator (50/60 Hz) ice-bath to create an emulsion. The silicone oil emulsion was diluted into a buffer solution within 10–15 min to avoid agglomeration and subsequent phase separation after completion of sonication. Over the duration period of the experiments, the resulting emulsion was verified to be stable in concentration and size distribution over time and with dilution. The silicone oil emulsion was diluted with PBS buffer at a concentration of 0.20% (w/v) and analyzed using MFI for particle size, concentration, and various morphological parameters. The sample was pretreated to remove the micro-air bubbles by letting the sample rest for approximately 30 min at 4 °C before analysis. Positive control samples containing only filtered PBS (without any silicone oil) were treated in similar manner and found to contain no air bubbles.

Protein particles. Monoclonal IgG1 antibody (1 mg/mL) was selected as the model protein for this study. Subvisible (and visible) protein aggregates were created with the antibody solution using a freeze-thaw method. The freeze-thaw method consisted of repeatedly freezing the protein sample at -80 °C for 5 min and immediately thawing it in a warm water bath at 37 °C for 5 min. The freeze–thaw cycle was performed five times after the original stock sample thaw. The count, size, and morphological characteristics of the protein particulates in the freeze–thaw treated antibody solution were analyzed using MFI. Positive control samples containing only filtered PBS (without protein), which were treated in a similar manner, were again verified to contain no air bubbles. The resulting concentration of protein particulates/aggregates was verified to be stable for the period of the experiments, as verified by time and dilution stability studies (data not included).

Combining silicone oil and protein particle solutions. The silicone oil emulsion and aggregated protein suspension were generated as outlined above. Protein and silicone oil stock solutions were combined in a 10-mL sample syringe to create concentrations of protein and silicone oil of 0.1 mg/mL and 0.20% (w/v) respectively. The samples were mixed by repeated, gentle pipetting and inspected visually to ensure a homogeneous appearance. Control samples of protein and silicone oil were prepared individually in PBS buffer by substituting the appropriate buffer for the protein and silicone-oil suspension. The test and control samples and their corresponding buffer blanks were prepared in quadruplet.

MFI measurements. Sample fluids were drawn, using a peristaltic pump, from a stirred (600 rpm) 10-mL sample syringe through a 400 μm flow-cell mounted in the HEV setup of an MFI instrument. Low shear forces on the fragile aggregates are ensured through low flow rates, gravity-assisted flow, uniform sample linear velocity, and minimized dead volumes. Sampling efficiency is maximized through flow cell design, material selection, and the use of hydrophobic coatings. Before each sample run, particle-free fluid was flushed through the system to achieve a clean baseline (0 particle counts per mL) and to optimize the illumination at the selected magnification. During the sample run, successive frames were displayed in grayscale or binary mode. These provide immediate visual feedback on the nature of the particle population as well as visual confirmation of the data obtained.

The measurements of particle size or morphology made by MFI are independent of the particle's material type. However, if all or part of a particle lacks sufficient contrast due to sufficiently high transparency, the corresponding particle may be undersized or missed entirely. To measure highly transparent particles such as silicone-oil droplets and highly transparent protein aggregates, the sensitivity of the instrument is designed to use very high threshold values, low noise electronics, and intelligent software algorithms.

Results

The experimental protocol was successful in creating repeatable and stable populations of subvisible and visible particles under controlled conditions. All samples were verified for time and dilution stability. The selected monoclonal IgG1 antibody readily forms aggregates and particulates under freeze-thaw conditions.

Figure 1: Silicone oil droplets in IgG1 antibody solution. (ALL FIGURES ARE COURTESY OF THE AUTHORS.)

Silicone microdroplet characterization. Sonication of the silicone oil sample resulted in the formation of an oil emulsion consisting of microdroplets of varying sizes ranging from 2 to 60 μm with the majority of these particles at approximately 5 μm. Positive control samples containing only filtered PBS buffer were analyzed for the presence of air bubbles and other contaminants. The PBS buffer sample was observed to be particle-free during the analysis, indicating that any air bubbles were removed during sample pretreatment. The homogenous population of siliconeoil droplets was observed to have a well defined circular shape with an aspect ratio >0.85 (see Figure 1). Experiments showed that the deviation from the ideal aspect ratio of 1.0, for circular objects, was the result of noise and pixilation effects at the high instrument sensitivity required to detect and measure the protein aggregates.

Figure 2: Generation of protein aggregates in IgG1 antibody solution. (ALL FIGURES ARE COURTESY OF THE AUTHORS.)

Protein particle generation and characterization. The untreated IgG1 formulation contained a small number of protein particles, believed to have formed during the first sample thawing. As shown in Figure 2, an eight-fold increase in the number of particle counts was observed by MFI in the antibody solution following five freeze–thaw cycles. The protein particles or aggregates were observed to be highly heterogeneous in shape, ranging from small dense fibers to large ribbon-like aggregates. Figure 3 contains representative images of protein particles with a wide range of values for the morphological parameters such as circularity, aspect ratio, and intensity. Air bubbles were not observed in the positive control samples. The results indicate that the freeze–thaw method is suitable for generating protein aggregates in this particular IgG1 formulation.

Figure 3: Images of typical protein particles. (ALL FIGURES ARE COURTESY OF THE AUTHORS.)

It was also observed that, with the increase in size of protein particles, both aspect ratio (see Figure 4) and mean intensity decreased (data not included).

Figure 4: Protein aggregates in IgG1 antibody solution. (ALL FIGURES ARE COURTESY OF THE AUTHORS.)

Differentiating silicone oil and protein particles. The mixed siliconeoil droplet and aggregated protein sample was measured using MFI. Visual analysis of the captured images showed that the two particle types could easily be resolved for particles ≥5 μm ECD. Because silicone-oil droplets have a consistently higher aspect ratio compared with aggregated protein particles of the same size (see Figure 5), a simple software filter with an aspect ratio ≥0.85 and ECD ≥5 μm cutoff was applied to the mixed population. The images obtained in the two individual populations prior to mixing were visually examined to assess the accuracy of this filter. This comparison showed an accuracy of 96% for pure samples (i.e., 4% of the particles were incorrectly labeled as either silicone oil or protein aggregates).

Figure 5: Mixed population of silicone-oil droplets and protein aggregates. (ALL FIGURES ARE COURTESY OF THE AUTHORS.)

Disscussion

Silicone-oil-induced particle formation in therapeutic proteins can be an issue for commercialization of the product. It is critical to characterize the nature of the particles resulting from the phenomenon of silicone-induced aggregation of proteins and antibodies. The results obtained by MFI analysis of mixed silicone-oil droplet and aggregate protein populations show that the two populations can be resolved with a high degree of accuracy using a simple software filter that uses aspect ratio and ECD limits. This level of accuracy would normally be sufficient when the concentration of the two particle types is comparable. If one population was much smaller than the other, then higher levels of accuracy could be achieved by including more morphological parameters in the analysis. If it was desired to extend the analysis to smaller particles, then the analysis could be carried out at higher magnification. These results clearly show the advantage of the MFI analysis over the light obscuration and filter-based techniques when attempting to isolate subpopulations.

The ability of MFI to resolve and independently measure populations of silicone-oil droplets and protein aggregates/particulates that are simultaneously present in heterogeneous samples can be used in a number of applications from packaging to formulation development. Some of these applications include:

• Orthogonal technique to other particle characterization techniques

• Selection of silicone type and application techniques to minimize droplet formation

• Qualification of silicone-oil microdroplets levels from coated containers

• Formulation optimization to minimize silicone-induced protein aggregation and particulation

• Development of advanced container or enclosures that do not shed silicone oil microdroplets into the protein formulation.

Conclusions

MFI with automated particle classification is an emerging technology that can play a useful role in understanding and controlling silicone-oil-droplet-induced aggregation of proteins in parenteral pharmaceuticals. More generally, the ability of the technology to resolve and independently characterize mixed particle populations, including a wide range of subvisible and visible particle types, offers a rapid and powerful means of evaluating subvisible and visible particle populations in parenteral products.

Deepak K. Sharma*, PhD, is a senior scientist in R&D, and Peter Oma is the director of R&D at Brightwell Technologies Inc., 115 Terence Matthews Crescent, Ottawa, Ontario, K2M 2B2, Canada, tel. 613.591.7715, fax 613.591.7716, dsharma@brightwelltech.com. Sampath Krishnan, PhD, is a senior scientist in Process & Product Development at Amgen Inc.

*To whom all correspondence should be addressed.

Submitted: July 21, 2008; Accepted: Aug. 28, 2008.

References

1. E.J. Smith et al., "Siliconization of Parenteral Drug Packaging Components," J. Parent. Sci. Technol. 42 (Supplement 4S), S1–S13 (1988).

2. S. Branley, "Chemistry and Properties of Medical Grade Silicones," J. Macromol. Sci. Chem A. 4 (3), 529–544 (1970).

3. E.A. Chantelau, "Silicone Oil Contamination of Insulin," Diabet. Med. 6 (3), 278 (1989). ]

4. R.K. Bernstein, "Clouding and Deactivation of Clear (Regular) Human Insulin: Association with Silicone Oil from Disposable Syringes," Diabetes Care 10 (6), 786–787 (1987).

5. R.N. Baldwin, "Contamination of Insulin by Silicone Oil: A Potential Hazard of Plastic Insulin Syringes," Diabet. Med. 5 (8), 789–790 (1988).

6. G.L. Shearer, "Contaminant Identification in Pharmaceutical Products," The Microscope 51 (1), 3–10 (2003).

7. L. S. Jones et al., "Silicone Oil Induced Aggregation of Proteins," J. Pharm. Sci. 94 (4), 918–927 (2005).

8. USP 29–NF 24 (suppl 2) General Chapter <788>, "Particulate Matter in Injections." US Pharmacopeial Convention, Rockville, MD, 2006.

9. "Particulate Contamination: Visible Particles" in European Pharmacopoeia, (European Pharmacopoeia Commission, Council of Europe, European Department for the Quality of Medicines, 2005) 5th ed. Vol. 5.0 (incl. suppl 5.1). General Chapter 2.9.20.

10. "Foreign Insoluble Particulate Matter Test for Injections" in Japanese Pharmacopoeia (Society of Japanese Pharmacopoeia, 2006) 14th edn. General Chapter 20.

11. D.K. Sharma et al., "Flow Microscopy for Particulate Analysis in Parenteral and Pharmaceutical Fluids," European J. Parent. Pharma. Sci. 12 (4), 97–101 (2007).