Flow Microscopy: Dynamic Image Analysis for Particle Counting

Transcription

1 STIMULI TO THE REVISION PROCESS Vol. 36(1) [Jan. Feb. 2010] of the USPC or the USP Council of Experts 311 Flow Microscopy: Dynamic Image Analysis for Particle Counting Peter Oma, Deepak K. Sharma, David King, Brightwell Technologies Inc., Ottawa, ON a ABSTRACT USP General Chapter Particulate Matter in Injections h788i sets limits and cites two enumeration procedures for subvisible particulate matter in parenteral products. Although h788i does not specifically target intrinsic sources of particulate matter such as protein aggregates, micelles, or precipitates, the h788i procedures can detect and enumerate these types of particles. The h788i procedure s performance (efficiency, reliability, and repeatability) for these particle types is unknown. Artifacts such as immiscible liquids, primarily silicone oil droplets, and air bubbles are counted by light obscuration but do not contribute to membrane microscopic counts. Protein-based pharmaceuticals may contain sizeable populations of aggregated or precipitated active ingredient in the size range of concern. The physical and optical properties of these particles can present challenges to light-obscuration procedures in achieving reliable detection and sizing. Because the light-obscuration h788i test procedures do not easily differentiate particle types, an alternative test procedure may be useful in differentiating foreign from intrinsic particulate matter to facilitate product improvement efforts and compliance with the limits specified by h788i. Flow microscopy is an imaging-based technology that uses automated classification algorithms to characterize suspended particle populations. The technology is currently used in an increasing number of applications in the formulation development phase of parenteral and biopharmaceutical development. The system operates by capturing an image of each particle in a flowing sample. Automated image analysis can differentiate subpopulations of foreign particles (e.g., metal, glass, rubber, and fibers) and intrinsic particles. This allows protein aggregates or other intrinsic particles to be considered separately from the total particle count to provide a more accurate representation of the foreign particle load. In addition, the particle images gathered by the system provide morphology information that can be used to determine the origin of contaminants and to monitor intrinsic particle populations. INTRODUCTION a Correspondence should be addressed to: Desmond Hunt, PhD, Scientist, USP, Twinbrook Parkway, Rockville, MD ; dgh@usp.org. Limiting the concentration of foreign particulate in parenteral pharmaceuticals is an essential requirement for patient safety (1). USP General Chapter Particulate Matter in Injections h788i and the harmonized versions in the European and Japanese pharmacopeias establish limits and cite procedures by which particle size and concentration should be measured. In biopharmaceuticals the tendency of protein molecules to form aggregates in the size range of concern poses a special challenge. Even though aggregates are not considered foreign material and do not lie within the scope of h788i, their presence can be incorrectly attributed to the foreign particle count of a particular formulation. Because these protein aggregates are at best active ingredient depleting and at worst may affect product safety with respect to inducing immunogenicity, measuring and limiting their concentration is also an important factor in ensuring patient safety. In addition to concerns about patient safety, scientists engaged in formulation development, stability testing, fill-and-finish processes, and lot-release testing have an interest in gaining further information about the characteristics of particulates in parenteral solutions. In formulation development and stability testing, knowledge of the concentration and morphological distributions of intrinsic particle subpopulations is required to maintain and guarantee product quality, especially for biopharmaceuticals. These intrinsic particles may be the drug active ingredient in native, aggregated, or precipitated form, micelles formed by polysorbate concentration, a lubricant such as silicone oil, or air bubbles caused by sample handling and/or preparation. When foreign particles are present, even within regulated limits, additional information about their nature and possible origin is valuable for process quality control and optimization. MEASURING PROTEIN AGGREGATES IN THE SUBVISIBLE SIZE RANGE The formation of protein aggregates can be induced by many factors, including processing stresses, storage conditions, concentration, temperature, and surface effects from contact materials such as glass and silicone (2). When protein aggregation cannot be eliminated by formulation design, populations that are present presumably should be controlled to consistent levels (size and concentration) and characteristics (morphology) during the formulation development, stability testing, clinical trials, manufacturing, and storage phases of the product life cycle. In a recent Commentary published in the Journal of Pharmaceutical Sciences, the authors reviewed the influence of populations of subvisible aggregates on the quality of therapeutic protein products (3). Among other conclusions, the article stressed the desirability of measuring particles sizes in the 0.1- to 10-mm range and the need to investigate new techniques for measuring aggregates and particles in the sub 10-mm range. Measurement procedures cited in h788i are automated light obscuration or, if this is not practical for the particular sample type, membrane microscopy. Both

2 312 STIMULI TO THE REVISION PROCESS of the USPC or the USP Council of Experts Vol. 36(1) [Jan. Feb. 2010] of these procedures work well for detecting and providing equivalent circular diameter (ECD) values for most particulate types. However, these procedures have limitations (described below) when more comprehensive characterization of challenging particle populations is required. Light obscuration is an indirect measurement procedure. When a particle transits the measurement zone of the instrument, an optical beam is obscured, with a resulting change in signal strength at the detector. This signal change is then equated to a particle s ECD based on a calibration curve created using polystyrene spheres of known sizes. When the particulates in real-world samples are not spherical or if they differ significantly from polystyrene in optical properties (e.g., opacity or refractive index), errors in sizing and counting may occur. Because protein aggregates are highly transparent, are nonspherical in morphology, and may possess a refractive index close to that of the buffer in which they are contained, they pose a particular challenge with respect to instrument sensitivity. The single ECD measurement provided for each particle by obscuration counting cannot provide information about the opacity, shape, nature, or origin of the particle. In addition, the obscuration calibration procedure depends on the relative material properties (e.g., opacity and shape) of polystyrene beads and thus does not allow accurate quantification of heterogeneous particle populations. Although membrane microscopy conducted by an experienced microscopist can provide detailed information, the technique is time consuming, is statistically limited, and does not easily provide quantitative results that are conveniently archived. Measurements can be produced using maximum chord, ECD, or Feret diameter, but all measurements must be conducted by comparison to a calibrated ocular graticule. Although visualizing actual particles has value and the sample can be preserved for further analysis, results obtained by observing and counting near-transparent particles can be subjective and can depend on the particular microscope configuration (microscope type, diaphragm setting, and illumination setup) and sample analysis protocol (filter medium color and sample transfer) and operator interpretation. These challenges have led to an interest in automated technologies that are less dependent on a particle s physical properties and can provide additional information about particle morphology (4,5). Flow microscopy is an automated, imaging-based particle-measurement technology designed for the analysis of particulates in pharmaceutical formulations. Imaging-based technologies offer a number of advantages by comparison with indirect light-obscuration or scattering techniques. They can have little dependence on the physical properties of a particle, can provide accurate sizing capability, and do not require user calibration. They also can be configured to have high sensitivity for detection of relatively transparent particles such as protein aggregates. Their image capture and analysis capabilities also provide quantitative and qualitative information about target populations and a digital archive of the results (6 8). FLOW MICROSCOPY TECHNOLOGIES A representative flow microscope instrument configuration is shown in Figure 1. Different commercial implementations of the technology are available, so specific details regarding imaging procedure, sample flow, magnification, and specifications will vary. However most flow microscopes have the same general mode of operation, including the capture of particle images as a sample stream passes through a flow cell centered in the field of view of a magnification system. In order to provide statistically meaningful results in practical timeframes, an enhanced depth-of-field procedure is usually employed. Once the images are acquired, they are analyzed by the system software to extract each particle s morphological parameters and to compile a database containing count, size, concentration, and intensity, as well as other shape parameters of interest such as area, perimeter, circularity, maximum Feret diameter, and aspect ratio. This database can be further interrogated by application software to produce parameter plots such as histograms and scatter plots. This software facility allows the user to visually select representative particles and to use these to automatically define parameter filters that can be used to categorically evaluate the data from the sample population. These capabilities allow particle subpopulations to be isolated and independently analyzed. Particle images can be archived and retrieved for visual confirmation, comparison, and further analysis. Depending on the specific system magnification chosen for the analysis, particle sizes can be analyzed in selected ranges from less than one micron to hundreds of microns. A flow microscope s ability to resolve and collect data for specific particle types depends on the particle s attributes, overall particle population statistics, and the instrument configuration.

3 STIMULI TO THE REVISION PROCESS Vol. 36(1) [Jan. Feb. 2010] of the USPC or the USP Council of Experts 313 Figure 1. Schematic Configuration of a Flow Microscope (CPU = central processing unit) (7). In the configuration shown, sample fluid is drawn from a sample container (syringe, beaker, pipette, etc.) through the flow cell using a metered pumping component. The pump is usually located on the output side of the flow cell to preserve the integrity of the sample. The flow cell design is critical to the successful operation of the instrument. For optimum performance during testing of parenteral products, flow microscopes should be configured for the unique requirements of these sample types. These requirements may include: low flow rates to minimize shear forces on fragile particles high sensitivity to detect nearly transparent particles (e.g., proteinaceous particles) gravity-assisted sample introduction minimal wetted components and minimized dead volumes ( ml is desirable) high sampling efficiency ability to analyze small volumes (approximately 0.3 ml) and low (approximately 5 per ml) particle concentrations. A variety of sample introduction procedures (pipettes, syringe barrels, tubing connections, and prefilled syringes) should be accommodated to handle the wide range of sample volumes and formulation types. QUALIFICATION FOR PARENTERAL APPLICATIONS For applications in the pharmaceutical environment, the performance of a new technology such as flow microscopy must meet or exceed the specifications required by established procedures. Table 1 lists a set of proposed tests to qualify and validate flow microscopy as a procedure complementary or supplemental to h788i. The sizing and concentration tests are similar to current Instrument Standardization Test (IST) procedures specified in h788i. The other tests in Table 1 verify the additional capabilities of flow microscopy. Testing with polystyrene reference beads to verify sizing and concentration accuracy should be part of any standard user protocol. The remaining tests should be carried out as part of an initial qualification procedure performed on a flow microscope based on the particular user s or manufacturer s application. Representative qualification data (size and concentration accuracy/range) for an instrument with a configuration appropriate for protein-based parenterals are shown in Figures 2A 2C. Table 1. Qualification Tests for Flow Microscopy Parameter Comparison to Light-obscuration IST Test Procedure Sizing Accuracy Same Measure bead size standards across the appropriate size range. Concentration Accuracy Same Measure bead concentration standards across the appropriate concentration range. Low Particle Concentrations New Measure bead concentrations < 10 particles/ ml. High Particle Concentrations New Measure bead concentrations > 100,000 particles/ml. Sensitivity New Measure beads of a known size (ECD) with a refractive index approaching that of the carrier fluid.

4 314 STIMULI TO THE REVISION PROCESS of the USPC or the USP Council of Experts Vol. 36(1) [Jan. Feb. 2010] Table 1. Qualification Tests for Flow Microscopy (Continued) Parameter Comparison to Light-obscuration IST Test Procedure Material Independence New Measure reference beads of known size (ECD) composed of different material types. Particle Shape Analysis New Measure particles with similar size (ECD) but different morphology (e.g., Feret diameter, circularity, or aspect ratio). Figure 2A: Flow Microscopy: Size Range Qualification (Line of Identity) (7). NIST = National Institute of Standards and Technology. Figure 2B. Flow Microscopy: Concentration Range Qualification (Line of Identity) (7).

5 STIMULI TO THE REVISION PROCESS Vol. 36(1) [Jan. Feb. 2010] of the USPC or the USP Council of Experts 315 Figure 2C. Flow Microscopy Results for USP Particle Count Reference Standard. COMPARISON TO h788i PROCEDURES For the purpose of comparing flow microscopy measurements with light obscuration and membrane microscopy, we measured an engineered solution containing proteinaceous particles. The parenteral formulation containing protein particles was provided by a pharmaceutical manufacturer. The engineered formulation was diluted with phosphate-buffered saline (ph ) containing sorbitol and polysorbate-80 provided by the manufacturer. The solution was prepared and analyzed using the h788i microscopic and obscuration procedures by an external laboratory that complied with current Good Laboratory Practices and current Good Manufacturing Practices. The same solution was also analyzed using a flow microscope, and the results of the three measurements are presented in Table 2 and Figure 3. Table 2. Comparison of Particle Concentration (7) Particles/mL Samples Flow Microscope Light Obscuration Microscopic Membrane 10 mm 25 mm 10 mm 25 mm 10 mm 25 mm BT BT BT

6 316 STIMULI TO THE REVISION PROCESS of the USPC or the USP Council of Experts Vol. 36(1) [Jan. Feb. 2010] Figure 3. Flow Microscopy and h788i Procedures (7). For particle sizes in the range of concern, the measured concentrations for this sample type differ by one or more orders of magnitude. This experiment has been repeated on different protein-based formulations in other pharmaceutical laboratories with similar results (7 9). Posttesting analysis of saved particle images confirms the higher concentrations measured by flow microscopy. The low count obtained by microscopic membrane probably is caused by a combination of factors including the modification (destruction or break-up) of aggregates due to the preparation protocol itself and operator interpretation of valid substances. The cause of the differences in results between flow microscopy and light obscuration is not fully understood. One of the contributing factors may be the differences in sensitivity between the two procedures when the refractive index of the target particles approaches that of the carrier fluid. This effect is seen in Figures 4A (obscuration) and 4B (flow microscopy) that show size and concentration measurements for a population of 4.80-mm, low refractive index glass beads (refractive index 1.43 to 1.46 at 589 nm) suspended in water and in a solution of 40% (v/v) ethylene glycol in water (refractive index approximately 1.37). Measurements are in agreement between the two techniques when the beads are suspended in 100% water. One notes undersizing for the obscuration measurement of the higher refractive index glycol mixture, but flow microscopy results are almost unaffected. Figure 4A: Obscuration Measurements of Low Refractive Index Glass Beads.

7 STIMULI TO THE REVISION PROCESS Vol. 36(1) [Jan. Feb. 2010] of the USPC or the USP Council of Experts 317 The refractive index of various protein aggregates is in the range of 1.33 to 1.40 (vs 1.59 for polystyrene bead standards), which contributes to the undersizing and undercounting detected with obscuration procedures. With flow microscopy, postanalysis of saved images can be used to determine whether particles are being undersized or fragmented because of limited instrument sensitivity. With proper choice of instrument configuration, these effects can be minimized for protein samples. Figure 4B. Flow Microscopy Measurement of Low Refractive Index Glass Beads. Although the difference in refractive index may be an important component, other factors could also contribute to the observed differences. These include: the differences in the optical absorption of proteins at different wavelengths (in this case the obscuration light source wavelength was 780 nm, and the flow microscope light source was 470 nm) impact of particle morphology on sizing particle fragmentation at higher shear forces induced by higher flow rates undetected coincidence effects in obscuration instruments particle settling or fragmentation due to magnetic stirring. EXAMPLE FLOW MICROSCOPY APPLICATIONS FOR PARENTERAL PRODUCTS Protein and Silicone Oil Isolation Silicone oil is used as a syringe and stopper lubricant. This oil can detach from surfaces and form droplets that may act as denaturing agents and nucleation centers in protein formulations (10). The detection and enumeration of silicone droplets are therefore important for product integrity and patient safety. With obscuration procedures alone it is difficult to characterize the relative distributions of silicone oil vs other foreign particles and protein aggregates present in the same sample. A flow microscope can detect and differentiate silicone oil microdroplets from other near-transparent particles (11). The results in Figure 5 are from a protein-based sample containing a controlled population of silicone oil microdroplets. A flow microscope was first used to measure the entire particle population. Subsequent analysis using the instrument s software-based morphology filters enabled the subpopulation of silicone oil to be identified and enumerated. In this experiment an aspect ratio 0.85 and ECD of 5 mm were chosen as the primary filters. The accuracy of these filters was manually verified by visual examination of stored images and comparison of the results with those provided by the instrument. In this example and using these parameters, the filter accuracy was 96% (i.e., 4% of the particles were incorrectly identified). During analysis, the application of additional morphological parameters such as circularity and intensity mean can further improve the accuracy of filters. The deviation from the ideal aspect ratio of 1.0 for the droplets is the result of practical image resolution limits and pixilation effects (quantization) for the particular configuration selected to detect and measure the highly transparent particles.

8 318 STIMULI TO THE REVISION PROCESS of the USPC or the USP Council of Experts Vol. 36(1) [Jan. Feb. 2010] Protein Aggregation Figure 5. Aspect Ratio Results for Protein and Silicone Oil Mixture. Characterizing and minimizing protein aggregation is a requirement in biopharmaceutical formulation development. Procedures such as size-exclusion chromatography and dynamic light scattering measure aggregates up to a size of approximately 500 nm. However, techniques to analyze low and high concentrations of protein aggregates from approximately 500 nm through the subvisible and visible regions (> 100 mm) are lacking (12, 13). Flow microscopy may be a suitable candidate to help expand the understanding of protein particle formation and, subsequently, to assist in controlling aggregation formation. The capabilities of flow microscopy are illustrated in the data shown in Table 3. These data were extracted from a study of the effects of stresses such as shaking, temperature, filtering, and aging in relation to their ability to induce protein aggregation (6). The results in Table 3 are from an experiment in which a model protein formulation IMGN901 (14) provided by ImmunoGen (Waltham, MA) at a concentration of 1 mg/ ml was subjected to shaking (Lab-line Model 3520 Orbital Shaker) at a rate of 200 rpm at room temperature for 24 h (7). One vial was held as a nonshaken control. At each time point, the number of particles ( 10 mm) in 5 ml of the sample was measured using flow microscopy. Once again the results for flow microscopy were verified by manual examination of stored images. Figure 6 contains representative particle images from the experiment. Table 3. Protein Particle Count for Shaken Sample Using Flow Microscopy Time (h) Counts/mL 0 12, , , , , , , (control at 0 rpm) 4785 Figure 6. Particle Images: A (0 h, 54 mm); B (4 h, 176 mm); C (6 h, 260 mm) (images not shown to scale).

9 STIMULI TO THE REVISION PROCESS Vol. 36(1) [Jan. Feb. 2010] of the USPC or the USP Council of Experts 319 Evaluation of Filtrate Liquids Filtering (normally 0.22 mm) is occasionally specified for parenteral products in order to remove particles in the h788i size range. However, the ability of the filters (including point-of-use filters) to eliminate undesirable protein particles is not clear (15, 16). To investigate the effects of filtration on the number and size distribution of protein particles, samples were measured pre- and postfiltration using flow microscopy. For comparison purposes, obscuration measurements were also conducted on the postfiltration samples. The obscuration measurements were not performed on the undiluted prefiltration sample because of the high particle concentrations that exceeded the instrument s operating range. Table 4 outlines the particle size and concentration measurement results for the protein filtration tests. Although filtration can substantially reduce the concentration of protein particles, it does not eliminate all protein particles because particles larger than the filter pore size of 0.22 mm canstillbefoundinthefilteredsamples.itisunknown whether the remaining protein particles in the postfiltration sample passed through the filter or were generated because of stability requirements in the solution. The filtered sample results measured by the two procedures show that flow microscopy measured a higher concentration of particles over the size range of interest. Table 4. Filtration Results Flow Microscopy Prefiltration Flow Microscopy Postfiltration Particle Size Obscuration Prefiltration a 2 10 mm 330, Not Done mm 60, Not Done mm 20, Not Done 50 a Undiluted particle concentrations exceed the specified limit of the obscuration instrument. CONCLUSIONS Flow microscopy with automated particle classification is an intelligent imaging technology that is finding increasing use in applications for evaluating populations of suspended particles encountered during formulation development of biopharmaceutical drugs. Compared to existing h788i procedures, flow microscopy seeks to combine the flexibility and visual verification of the manual microscopy procedure with the speed, statistical accuracy, and quantification of the light-obscuration procedure. The technique also adds capabilities, including increased sensitivity for transparent particles such as protein aggregates, higher concentration limits, particle morphology analysis and classification, material independence, data archive for all image and morphology parameters, and an ability to isolate particle subpopulations of interest. The combined image and particle information provided by flow microscopy can be employed in a variety of applications in the parenteral development and manufacturing environments. Examples include formulation development, stability testing, contaminant isolation, process control, quality control, diagnostics, and troubleshooting. These additional capabilities may justify consideration of flow microscopy as a complement to existing h788i test procedures. REFERENCES 1. Borchert S, Abe A, Aldrich S, Fox L, Freeman J. Particulate matter in parenteral products: a review. J Parenter Sci Technol. 1986;40(5): Mahler H-C, Friess W, Grauschopf U, Kiese S. Protein aggregation: pathways, induction factors and analysis. J Pharm Sci. 2008, published online. DOI: /jps Obscuration Postfiltration 3. Carpenter J, Randolph T, Jiskoot W, et al. Overlooking subvisible particles in therapeutic protein products: gaps that may compromise product quality. J Pharm Sci. 2008, published online. DOI: /jps Burgess DJ, Duffy E, Etzler F, Hickey AJ. Particle size analysis: AAPS workshop report. AAPS J. 2004;6(3):article Ives C, Soderquist R, Stoner M, Kendrick B. Light obscuration particulate analysis for protein solutions: challenges and limitations. Colorado Protein Stability Conference. 2007, Breckenridge, CO, USA. 6. Krishnamurthy R, Sukumar M, Das T, Lacher N. Emerging analytical technologies for biotherapeutics development. Bioprocess Intl. 2008;6(5): Huang C, Sharma D, Oma P, Krishnamurthy R. Quantitation of protein particles in parenteral solutions using Micro-Flow Imaging. JPharmSci.2008, published online. DOI: /jps Sharma D, King D, Moore P, Oma P, Thomas D. Flow microscopy for particulate analysis in parenteral and pharmaceutical fluids. Eur J Parenter Pharm Sci. 2007;12(4): Huang C, Sharma D, Amplett G, Oma P, Krishnamurthy K. Quantitation of protein particles in parenteral solutions using light obscuration and micro flow imaging, Colorado Protein Stability Conference. 2007, Breckenridge, CO, USA. 10. Jones L, Kaufmann A, Middaugh C. Silicone oil induced aggregation of proteins. J Pharm Sci. 2005;94: Sharma D, Oma P, Krishnan S. Silicone micro-droplets in protein formulations detection and enumeration. Pharm Technol. 2009;33(4). 12. Philo, JS. Is any measurement method optimal for all aggregate sizes and types? AAPS J. 2006;8(3):article Cromwell M., Hilario E, Jacobson F. Protein aggregation and bioprocessing. AAPS J. 2006;8(3):article 66.

10 320 STIMULI TO THE REVISION PROCESS of the USPC or the USP Council of Experts Vol. 36(1) [Jan. Feb. 2010] 14. IMGN901. Available at: page/imgn901b. Accessed on July Ernerot L, Sandell E. Membrane filtration during administration of infusion fluids for elimination of particulate matter. Acta Pharm Suecica [Acta Pharm Nordica]. 1967;4(5): Turco SJ, Davis NM. Comparison of final filtration devices. Bull Parenter Drug Assoc. 1973;27(5):