Practical Considerations for Detection and Characterization of Sub-Micron Particles in Protein Solutions by Nanoparticle Tracking Analysis.

Downloaded from journal.pda.org on October 7, 2015

Practical Considerations for Detection and Characterization of Sub-Micron Particles in Protein Solutions by Nanoparticle Tracking Analysis Flaviu Gruia, Arun Parupudi and Alla Polozova

PDA J Pharm Sci and Tech 2015, 69 427-439 Access the most recent version at doi:10.5731/pdajpst.2015.01051


TECHNOLOGY/APPLICATION

Practical Considerations for Detection and Characterization of Sub-Micron Particles in Protein Solutions by Nanoparticle Tracking Analysis FLAVIU GRUIA1,*, ARUN PARUPUDI1, and ALLA POLOZOVA2 1

Department of Analytical Biotechnology, MedImmune, Gaithersburg MD 20878; and 2Analytical Sciences, Process Development, Amgen, West Greenwich, RI 02817 ©PDA, Inc. 2015 ABSTRACT: Nanoparticle Tracking Analysis (NTA) is an emerging analytical technique developed for detection, sizing, and counting of sub-micron particles in liquid media. Its feasibility for use in biopharmaceutical development was evaluated with particle standards and recombinant protein solutions. Measurements of aqueous suspensions of NIST-traceable polystyrene particle standards showed accurate particle concentration detection between 2 ⫻ 107 and 5 ⫻ 109 particles/mL. Sizing was accurate for particle standards up to 200 nm. Smaller than nominal value sizes were detected by NTA for the 300 –900 nm particles. Measurements of protein solutions showed that NTA performance is solution-specific. Reduced sensitivity, especially in opalescent solutions, was observed. Measurements in such solutions may require sample dilution; however, common sample manipulations, such as dilution and filtration, may result in particle formation. Dilution and filtration case studies are presented to further illustrate such behavior. To benchmark general performance, NTA was compared against asymmetric flow field flow fractionation coupled with multi-angle light scattering (aF4-MALS) and dynamic light scattering, which are other techniques for sub-micron particles. Data shows that all three methods have limitations and may not work equally well under certain conditions. Nevertheless, the ability of NTA to directly detect and count sub-micron particles is a feature not matched by aF4-MALS or dynamic light scattering. LAY ABSTRACT: Thorough characterization of particulate matter present in protein therapeutics is limited by the lack of analytical methods for particles in the sub-micron size range. Emerging techniques are being developed to bridge this analytical gap. In this study, Nanoparticle Tracking Analysis is evaluated as a potential tool for biologics development. Our results indicate that method performance is molecule-specific and may not work as well under all solution conditions, especially when testing opalescent solutions. Advantages and disadvantages of Nanoparticle Tracking Analysis are discussed in comparison to other analytical techniques for particles in the sub-micron size range.

Introduction The presence of particles in therapeutic protein products and their potential impact on product quality have been the subject of a considerable number of publications. Particle formation may be triggered by various protein-specific characteristics as well as extrinsic variables during processing, storage, and/or adminis-

*Corresponding Author: Flaviu Gruia, Department of Analytical Biotechnology, MedImmune, One Medimmune Way, Gaithersburg MD 20878. E-mail: [email protected] doi: 10.5731/pdajpst.2015.01051

Vol. 69, No. 3, May–June 2015

tration (1– 4). The potential for protein particles to cause immunogenic responses in patients is one of the main reasons behind the increased attention to particle analysis in therapeutic protein products (5–7). As a result of this increased awareness, new technologies and instrumentation are being developed to improve particle detection and characterization (8 –11). One example of this is Nanoparticle Tracking Analysis (NTA) (12, 13). NTA has been specifically developed for the assessment of sub-micron particles in solution. The underlying technology is a combination of light scattering, microscopy, and particle tracking software. Particles are detected based on their differential light scattering with respect to the bulk solution. The concentration of particles is derived from extrapolating the number of detected scattering centers in the im427


aged volume. Individual particle size information is computed by tracking the diffusion of each detected particle. Although not exclusively developed for measurements of sub-micron particles in protein solutions (14 –16), NTA has already been employed in the analysis of protein therapeutics and vaccines (1, 17). This study evaluates the applicability of NTA for detection and analysis of sub-micron populations of particles in recombinant protein (specifically, monoclonal antibodies [MAbs]) solutions. To establish a baseline dynamic range, linearity, size accuracy, and sensitivity, a series of measurements using polystyrene particle standards of known size and concentrations was carried out. The method was further tested with a range of protein solutions to assess whether its performance differed from that of non-proteinaceous particle standards. The effects of common sample handling operations, such as dilution and filtration, were measured for three MAb solutions to evaluate the potential impact of such operations on sub-micron particle populations present in protein solutions. Additionally, a direct comparison against other techniques such as dynamic light scattering (DLS) (10, 18, 19) and asymmetric flow field flow fractionation coupled with multi-angle light scattering (aF4-MALS) (20 –22) was performed with the specific objective of independently benchmarking NTA. The advantages and disadvantages of NTA are discussed in the context of specific samples. Materials and Methods Particle Standards Aqueous suspensions of polystyrene particle standards with a nominal diameter of 125 nm were purchased from Duke Scientific (Palo Alto, CA). Solutions with particle concentrations ranging from 2 ⫻ 106 to 1010 particles/mL were generated by serial dilution with distilled ultrapure water and used for range and concentration linearity assessments.

used to test whether photophoresis interfered with size accuracy measurements. Protein Solutions Protein solutions of MAb 1 (MAb1), MAb 2 (MAb2), and MAb 3 (MAb3) were provided by MedImmune (Gaithersburg, MD). Samples of varying concentrations were generated through dilution with specific formulation buffers. MAb1-stressed samples were generated by mild mechanical stress (12 h slow stirring with a Teflon-coated magnetic stirrer). A portion of the stressed MAb1 solution was filtered through a 450 nm pore size filter and used for comparative evaluation of NTA, DLS, and aF4-MALS. All protein solutions and dilution buffers were degassed under low vacuum (4 psi) prior to sub-micron particle measurements. NTA Measurements Sub-micron particle measurements were performed with a NanoSight LM17C instrument (NanoSight, Amesbury, UK) equipped with a 532 nm diode laser, using the proprietary NTA 2.1 built 03.09 software version for data collection and analysis. Two microscope objectives (10⫻ and 20⫻ magnification) were employed to record the video files for NTA analysis. The typical volume of sample used for each measurement was 0.5 mL. Recording parameters (shutter, gain) were optimized for each sample and kept constant while collecting five to eight replicates (30 seconds/replicate) for each sample. The data analysis was performed with automatic particle detection and optimized, sample-specific, analysis parameters (brightness, gain, and blur). The corresponding output files were further processed in Origin 8 (Microcal, Northampton, MA) to compute average particle concentrations and size distributions, evaluate the experimental variability, and generate plots. DLS Measurements

Aqueous suspensions of polystyrene particle standards with nominal diameters of 30, 60, 100, 200, 300, 400, and 900 nm were purchased from Thermo Scientific (Waltham, MA). Solutions with particle concentrations ranging from 108 to 109 particles/mL were generated by serial dilutions in distilled ultrapure water from the stock solutions and used for size accuracy measurements. A 5:1 mixture of 100 nm and 300 nm polystyrene particle standards was also prepared and 428

DLS measurements were carried out with a Wyatt DynaPro plate reader (Santa Barbara, CA) using the proprietary Dynamics software, version 7.1.7. Eight replicates (45 ␮L per replicate) were run for each sample, and autocorrelation functions were recorded for individual replicates. Data were further processed using the regularization algorithm to retrieve size distributions as a function of total scattering volume in PDA Journal of Pharmaceutical Science and Technology


solution. Results from the eight replicates were averaged to compute the final DLS size distributions. aF4-MALS Measurements aF4-MALS experiments were carried out using an Eclipse-3 system (Wyatt Technologies, Santa Barbara, CA) equipped with a trapezoidal channel with a 10 kDa regenerated cellulose membrane, serially connected to a DAWN HELEOS-II (Wyatt Technologies, Santa Barbara, CA) and an Agilent UV detector. The mobile phase used for elution was 0.1 M sodium sulfate, 0.1 M di-basic sodium phosphate, pH 7.2. The samples were diluted to 1 mg/mL with the MAb1 formulation buffer, and 15 ␮L of each sample was injected using an Agilent autosampler system. The samples were focused at 3 mL/min for 2 minutes and eluted with an initial cross flow of 2.0 mL/min for 10 min. The cross-flow was gradually decreased to 0.0 mL/min over a 10 minutes time interval. The elution continued for another 10 minutes without any crossflow. The elution flow rate was kept constant at 1 mL/min. Geometric radius was reported by fitting angular dependence of light scattering signals collected at 32, 38, 44, 50, 57, 64, 72, 81, and 90 degrees and extrapolating to zero angle using the Astra V software. Turbidity Measurements Turbidity was measured using a Hach (Loveland, CO) 2100AN Laboratory Turbidimeter equipped with 13 mm test tube adaptors. The instrument was calibrated with Formazin turbidity standards. A volume of 3 mL of each sample was transferred into 13 mm clear glass test tubes and turbidity was measured directly without further manipulations. The results were reported in nephelometric turbidity units (NTU). Spike-In Study Design Spike-in protocols were designed to test the method performance under different concentrations of protein or particles in solution. One protocol was designed to test the effect of the protein matrix on detection efficiency. Polystyrene microspheres with a nominal diameter of 125 nm were spiked into MAb1 solutions (low turbidity) with protein concentrations varying from 0 to 50 mg/mL. The final concentration of polystyrene microspheres was 9 ⫻ 107 particles/mL for all samples. A similar apVol. 69, No. 3, May–June 2015

proach was employed to investigate the effect of high– background signal solutions on particle detection. MAb2 solutions ranging from 0 to 50 mg/mL and having increased opalescence, as determined by nephelometry, were used in this case, and the theoretical particle concentration for all samples was 4 ⫻ 108 particles/mL. Another protocol was designed to evaluate the effect of particle concentration on detection sensitivity. In this case, samples with constant protein concentration (MAb1, 10 mg/mL) were used while the particle concentration was varied between 6 ⫻ 106 and 4 ⫻ 108 particles/mL. Dilution Study Design Solutions of MAb1, MAb2, and MAb3 at 10 mg/mL were prepared by dilution with corresponding formulation buffers from their original concentrations of 100, 100, and 150 mg/mL, respectively. NTA submicron particle evaluation for the diluted samples was executed immediately after dilution and then 2, 4, 7, and 24 h after dilution. All samples were kept at 2– 8 °C between measurements. Filtration Study Design Solutions of MAb1, MAb2, and MAb3 at 10 mg/mL were filtered through identical 0.22 ␮m filters (Millex GV Durapore polyvinylidene difluoride [PVDF] membrane, Millipore, Billerica, MA). Sub-micron particle measurements were performed immediately after filtration and then 2, 4, 7, and 24 h after filtration. All samples were kept at 2– 8 °C between measurements. Results and Discussion Assessment of the NTA Dynamic Range, Linearity, and Size Accuracy with Sub-Micron Particle Standard Suspensions Assessment of method range and concentration linearity was carried out with suspensions of 125 nm polystyrene microspheres (NIST-traceable). The results are summarized in Figure 1. Particle concentrations in the test solutions ranged from 2 ⫻ 106 to 1010 particles/mL. Measurements between 2 ⫻ 106 and 2 ⫻ 109 particles/mL used a 10⫻ magnification objective for data collection (red star series). A 20⫻ magnification objective was employed for measuring samples with expected particle concentrations of 108 to 1010 parti429


uncertainties at the lower end of the concentration range. For example, a change from 2 to 3 particles in the sampling volume would result in a 50% increase in the reported particle concentration. At low particle concentrations there could also be a significant interference from the background signal. Based on historical data, background signal is generally ⬃5 ⫻ 106 particles/mL.

Figure 1 NTA concentration range and linearity assessment with aqueous suspensions of 125 nm polystyrene beads. A 10ⴛ microscope objective was employed for measurements between 2 ⴛ 106 and 2 ⴛ 109 (red star series). A 20ⴛ microscope objective was employed for measurements at higher particles concentration, between 108 and 1010 (blue diamond series). Standard deviation of the measurements is shown as error bars for each measured concentration. The expected particle concentrations, based on dilution factors, along with error bars due to pipetting inaccuracies, are shown as the dashed black line. The grey area represents the suggested range of sub-micron particle concentrations that can be evaluated through NTA measurements.

cles/mL (blue diamond series). The dashed line corresponds to the expected concentration of particle standards based on dilution factors. Pipetting errors of up to 3% were assessed gravimetrically. Accordingly, error bars to account for the propagating effect of such pipetting inaccuracies on particle concentrations are shown. A better that 20% accuracy was observed for concentrations between 2 ⫻ 107 and 5 ⫻ 109 (shaded region) particles/mL. An accuracy of better than 10% was determined for particle concentrations in the range from 108 to 3 ⫻ 109 particles/mL. Deviations from linearity were detected at both the lower and the upper ends of the tested particle concentrations. Higher-than-expected particle counts were detected at the lower end of the concentration range (below 107 particles/mL). Statistically speaking, fluctuations in the number of particles may cause large 430

At the upper end of the concentration range the opposite effect was observed. Starting from a concentration of about 109 particles/mL (using the 10⫻ magnification objective) or 5 ⫻ 109 particles/mL (using the 20⫻ magnification objective) there was a marked decrease in the measured particle concentrations compared to the expected concentrations. This effect is likely due to the software limitations (i.e., inability to follow all the scattering centers, if more than ⬃200 scattering centers are present in each image, or inability to differentiate two scattering centers, if too close). Therefore, samples with particle concentrations approaching or above 5 ⫻ 109 particles/mL will require additional dilution for accurate sub-micron particle concentration assessment. Size accuracy was evaluated by measuring suspensions of NIST-traceable particle standards (particle concentration in the range 108–109) with nominal diameters between 30 nm and 900 nm. The results are summarized in Figure 2. Excellent sizing accuracy was recorded for particle standards with diameters between 30 and 200 nm. For solutions of particle standards with larger diameters (300, 400, and 900 nm), the size calculated by NTA was lower than the listed standard size by as much as 20%. The differences between the expected and the experimental values were significantly larger than measurement variability. A possible explanation for such a result could be the phenomenon of photophoresis (11, 23). Momentum transfer from the laser beam photons to the particle standards, due to the large difference in the index of refraction between the polystyrene and water bulk, could cause a faster diffusion of a sub-micron particle standards, with a stronger component along the laser beam direction. As a consequence, a smaller size will be assigned by the NTA software. Considering that photophoretic velocity is size-dependent, the effect should be more apparent for larger particles (24). This would manifest itself as an apparent flow of the particles along the laser beam direction. The NTA software, in principle, should be able to account for PDA Journal of Pharmaceutical Science and Technology


Figure 2 NTA size measurement of NIST-traceable particle standards with certified diameters of 30, 60, 100, 200, 300, 400, and 900 nm (red dot series) compared against their nominal sizes (black dashed line). The measured sizes of 100 nm and 300 nm particle standards as measured in a mixture of 100 nm and 300 nm particles standards are shown as green squares.

and minimize flow effects in the measured samples by identifying a drift velocity, which can be interpreted as the particles’ directional flow velocity. Drift velocity effects on free diffusion are deconvoluted and removed when computing particle sizes. However, the NTA software computes a single drift velocity. If photophoresis occurs, particles of different sizes would be subjected to different momentum transfer from the laser beam. Therefore, such correction will not be valid for all particles within broad size distributions. To further investigate the photophoresis hypothesis, an aqueous solution containing a 5:1 mixture of 100 nm and 300 nm particle standards was evaluated. The measured particles sizes under these conditions were 101 ⫾ 2 nm and 247 ⫾ 4 nm, respectively, slightly different than results obtained when individual particle standards were evaluated, 99 ⫾ 3 nm and 262 ⫾ 6 nm, respectively. The sizes measured in the mixed sample seem to support the idea that photophoresis effects may not be fully corrected by measuring an average drift velocity for the solution. Drift velocity correction would work better for the more numerous size populations present in solution (in this case the 100 nm particle). The differential speed acquired by Vol. 69, No. 3, May–June 2015

larger particles (300 nm particles in this case) will not be properly accounted for by the NTA correction algorithm. The measured size of the 300 nm particles in the mixture sample was 247 ⫾ 4 nm, which is lower than the 262 ⫾ 6 nm value obtained for the monodisperse 300 nm particle suspensions. This result is consistent with the occurrence of photophoresis. For the mixture of 100 nm and 300 nm beads, the common drift velocity assigned to all particles will likely be estimated from and skewed toward the major population of particles (100 nm). That drift velocity correction will not fully compensate for the higher level of momentum transfer to the 300 nm particles. Therefore, the 300 nm particles will be recorded as faster moving and a smaller size will be assigned accordingly. It is worth noting that NTA size accuracy measurements have been carried out in previous studies (13) and good correlation with nominal values of the particle standards were observed, including for mixtures of different size particles. However, a different laser source was employed in that instrument (640 nm wavelength versus 532 nm in current study), which may explain why measurable size discrepancies were observed here. Although photophoresis may produce measurable effects in aqueous solutions of particle standards due to large difference in index of refraction (⌬n ⫽ 0.265 for a polystyrene–water pair, at 532 nm), such effects will not be as pronounced when measuring protein particles in protein solutions. Assuming a 100 mg/mL protein solution with protein particles density of 600 mg/mL, the difference in the index of refraction between the solution and the particles will be ⌬n ⬍ 0.1. As the difference in index of refraction between the protein particles and their background protein solution is smaller (25), the momentum transfer to particles should diminish and therefore the sizing should be more accurate. The Effect of Optical Properties of Protein Solutions on NTA Sensitivity and Accuracy The protein matrix effect on sensitivity was evaluated at constant spike of particle standards while the protein concentration (MAb1 and MAb2) was varied between 0 and 50 mg/mL. The average detected particle concentrations along with their corresponding solution turbidity are summarized in Figure 3 for MAb1 and Figure 4A for MAb2. 431


Figure 3 Detection efficiency of a constant spike-in of 125 nm polystyrene beads in MAb1 solutions of varying concentrations (0 –50 mg/mL) and turbidity.

At low MAb1 concentrations (below 10 mg/mL) the measured particle concentrations were consistent with the expected theoretical values, corresponding to 100% (within the experimental error) spike-in recovery. However, there was a gradual decrease in spiked particles recovery to ⬃80% as the protein concentration was further increased to 50 mg/mL and the turbidity went up to 5.3 NTU. These results suggest an inverse correlation between protein concentration and method sensitivity (lower sensitivity at higher protein concentrations). One possible cause for the decrease in

Figure 4A Detection efficiency of a constant spike-in of 125 nm polystyrene beads in MAb2 solutions of varying concentrations (0 –50 mg/mL) and turbidity. 432

Figure 4B The effect of solution opalescence on NTA sensitivity: frame grabs from measurements of 125 nm polystyrene beads in MAb2 solutions of increasing protein concentration and turbidity.

sensitivity could be a reduction in contrast between the particles and the surrounding solution due to increased background light scattering from the solution matrix. To further investigate this hypothesis, a similar spike in study, using MAb2 instead, was executed. As opposed to MAb1, MAb2 solutions appear slightly opalescent at 50 mg/mL and show a steeper turbidity dependence on protein concentration. A more comprehensive description of the optical properties of MAb2 was provided in a previous study (26). For MAb2 solutions, the NTA particle detection efficiency dropped precipitously, especially at higher protein concentrations, as shown in Fig 4A. In the 50 mg/mL protein solution less than 10% of the spiked-in polystyrene particles were detected. A graphical description of how sample matrix affects particle detection is shown in Figure 4B. Increased background scattering due to opalescence makes visualization of particles very difficult. These studies show that NTA sensitivity PDA Journal of Pharmaceutical Science and Technology


Figure 5 Spiked-in recovery of varying concentrations (4 ⴛ 106 to 4 ⴛ 108 particles/mL) of 125 nm polystyrene bead spikes in Mab1 solutions of constant protein concentration (10 mg/mL). The 80 –100% recovery interval is shown by the dashed lines.

can be strongly affected by the solution matrix. In some cases, like concentrated MAb2 solutions, measurements are not possible unless samples are diluted. The effect of particle concentration on the accuracy of NTA detection in the protein matrix solution was assessed by keeping MAb1 concentration constant at 10 mg/mL while changing the concentration of spiked-in polystyrene microspheres. The summary of the experimental findings under these conditions is presented in Figure 5. For the range of 107 to 5 ⫻ 108 particles/mL, the spike-in recoveries were generally consistent (80 –120%) with the theoretically expected values. For samples with spiked-in polystyrene beads below 107 particles/mL, overestimation of sub-micron particle concentration was observed, matching results presented in Figure 1. These results may be explained by fluctuations of particle numbers in the sampling volume, which at the lower end of particles concentration can result in large deviations. NTA Performance with Protein Particles Particle-containing solutions of two MAb molecules were used to assess the capabilities of NTA for analysis of protein solutions. Protein particles were either generated under specific stress conditions (MAb1) or under normal processing and storage conditions (MAb2). Vol. 69, No. 3, May–June 2015

Figure 6A The effect of protein concentration on NTA measurements: Sub-micron particle counts in MAb1 solutions show a monotonic increase with protein concentration over the full range of measured concentrations. The inset shows a frame grab for the 100 mg/mL, 7 NTU sample. Particles are easily identified and counted by the NTA software.

Sub-micron particle measurements were carried out in MAb1 samples of variable protein concentration prepared by serial dilutions from a stock of 100 mg/mL stressed MAb1 solution. The average sub-micron particle counts from at least five individual measurements were computed and plotted against the protein concentration (Figure 6A). The particle counts are monotonically increasing with increased protein concentration. Although monotonically increasing, the relationship between submicron particle counts and protein concentration was not linear. NTA measurements of intrinsic sub-micron particle counts in MAb2 solutions, generated and assessed under conditions similar to MAb1, are shown in Figure 6B. The relationship between the particle counts and protein concentration was highly non-linear. For samples with protein concentration lower than 10 mg/mL there was a monotonic increase in particle counts with increasing protein concentration. However, above 10 mg/mL protein, there was a marked leveling off in the measured particle counts. These results confirm that detection of protein particles may be greatly affected by solution optical properties like turbidity and/or opalescence, not necessar433


performed for each situation, and decisions on what dilution factor to be used need be made on a case-bycase basis. Nevertheless, as detailed later in this study, dilution may sometimes affect particle counts, which may add additional complications to such measurements. Impact of Sample Handling on Sub-Micron Particle Distributions

Figure 6B The effect of protein concentration on NTA measurements: Sub-micron particle counts in MAb2 solutions show a nonlinear response with respect to protein concentration. The inset shows a frame grab for the 100 mg/mL, 47 NTU sample. Particles are hardly detectable.

ily protein concentration. MAb1 solution was clear, and it showed reduced light scattering background even at highest protein concentration (7 NTU at 100 mg/mL; see inset in Figure 6A). Sub-micron particles present in MAb1 solutions were readily identified and counted by NTA over the entire 0 –100 mg/mL concentration range. For MAb2 samples the NTA performance at higher concentrations was poor. The quality of the raw data was significantly reduced, and the increased background signal forced adjustments of the sensitivity settings, further increasing the contribution of the background noise. At concentrations of 100 mg/mL (47 NTU; see inset Figure 6B), sub-micron particles were difficult to identify from the background. From an analytical perspective, method performance dependence on solution optical properties makes development of a high-throughput or platform method rather difficult, as extensive solution characterization needs to be carried out before sub-micron particles measurements. Ideally, NTA measurements of a specific solution should be performed in a range where the method performs linearly, and then the sub-micron particle concentrations results must be extrapolated to the original solution conditions. Particle load versus protein concentration linearity assessment should be 434

Particle counts measurements, especially in the low micron and sub-micron size ranges, are often highly variable and inconsistent (27). The reproducibility of particle measurements in protein solutions might be affected by sample manipulations that are not generally identified as common causes for such effects (3, 4). In this study we have focused on dilution and filtration, which are often used in routine sample preparations. Although there is a considerable range of possible manipulations that samples may be subjected to during bioprocessing, we specifically explored the impact of dilution and filtration due to their ubiquity in standard laboratory procedures. Based on historical data, the experiments designed to test the effect of sample handling have been carried out at a protein concentration of ⬃10 mg/mL, to rule out potential turbidity interference. To evaluate the impact of dilution on sub-micron particle populations, particle counts were monitored in MAb1, MAb2, and MAb3 solutions immediately after dilution to 10 mg/mL from 100 mg/mL or 150 mg/mL stock. The time evolution of sub-micron particle counts during the 24 h after dilution is shown in Figure 7A. The trends in particle counts in the three tested MAb samples appear to follow different patterns. While the particle counts in MAb1 sample were virtually unchanged within the experimental variability during the 24 h period of the study, MAb2 and MAb3 solutions showed significant increases in the measured particle concentration, which peaked as soon as 2 h after dilution. The particle counts in MAb3 sample increased from 0.14 ⫻ 108 to 0.43 ⫻ 108 particles/mL. For MAb2 samples the counts went from 0.16 ⫻ 108 to 0.36 ⫻ 108 particles/mL. Although trending down, the particle counts in MAb2 and MAb3 samples remained elevated over the duration of the study and did not return to the original levels within the 24 h period. To assess the effect of filtration on the sub-micron particles, MAb1, MAb2, and MAb3 solutions at 10 mg/mL were filtered through a 0.22 ␮m PVDF filter PDA Journal of Pharmaceutical Science and Technology


Figure 7A Sub-micron particle trending in 10 mg/mL solutions of MAb1, MAb2, and MAb3 after dilution with corresponding formulation buffers. Measurements were carried out immediately after dilution and then 2, 4, 7, and 24 h after dilution. and monitored for 24 h. The experimental results are shown in Figure 7B. The measured particle concentrations were constant within the experimental error (0.10 ⫾ 0.03) ⫻ 108 particles/mL for MAb1 solutions. A measurable impact of filtration on sub-micron particles was observed for the MAb2 sample. The particle

Figure 7B Sub-micron particles in 10 mg/mL solutions of MAb1, MAb2, and MAb3 after 0.22 ␮m filtration (PVDF membrane). Measurements were carried out before filtration, at time 0 (immediately after filtration) and then 2, 4, 7, and 24 h after filtration. Vol. 69, No. 3, May–June 2015

counts increased from 0.24 ⫻ 108 to 0.93 ⫻ 108 particles/mL immediately after the filtration procedure. Within the next 2 h they dropped back to 0.48 ⫻ 108 particles/mL and trended lower for the remainder of the time points, with a value of 0.34 ⫻ 108 particles/mL recorded 24 h after filtration. Sub-micron particle counts in MAb3 sample also rapidly increased soon after filtration: from 0.22 ⫻ 108 to 0.53 ⫻ 108 particles/mL and then gradually decreased and leveled off to 0.29 ⫻ 108 particles/mL within the 24 h of the study. Although air bubble formation can be a potential source of the increase in sub-micron particle counts, we consider it unlikely. All protein samples and buffers were degassed prior to the beginning of the study, the viscosities were low and comparable, and the samples were gently handled. Different responses of MAb1, MAb2, and MAb3 solutions to the dilution suggest that properties of individual molecules and formulations affect the interactions responsible for the particle formation. Therefore, particle assessment in protein solutions should be carried out, as much as possible, in undiluted conditions. If that is not an option, dilution effects need to be investigated and understood. Similar observations were made for the filtration experiment. Although all three MAb samples, having the same protein concentrations and similar initial particle counts, were subjected to identical manipulations, the sub-micron particle formation trends were very different. MAb2 and MAb3 samples were extremely sensitive to this type of stress and showed significant increases in particle counts upon filtration. In contrast to this, no significant change in particle counts was observed for the MAb1 solution. One difference from the particle formation triggered by dilution is shorter time needed for the development of peak particle counts. Although there was a gradual reduction in the measured counts over the duration of the study, the final counts were still higher than at the beginning of the study. Interestingly, flow microscopy measurements carried out on MAb2 and MAb3 samples after dilution (data not shown) showed different dynamics in the subvisible range, namely a monotonic increase of larger than 4 ␮m particle counts over the whole time interval considered in this study. Subsequent analysis of the NTA sub-micron particle distributions showed that the size of particle distributions, as measured by D50, trended higher for duration of the study (from 73 nm 435


Figure 9 Figure 8 Overlay of particle size distributions measured by NTA (green) and DLS (burgundy) in a 2.5 mg/mL solution of stressed MAb1.

to 158 nm for MAb2, and from 104 nm to 360 nm, for MAb3, respectively). Such data trends support a particle formation mechanism where smaller sub-micron particles formed immediately after dilution further aggregate to form micron size or larger sub-visible particles. This model can reconcile the gradual reduction in sub-micron particles counts coincidental with an increase in their average sizes and an increase in the larger, micron-size particle counts. Even though the exact mechanisms behind the type of behavior observed for MAb2 and MAb3 solutions are not fully understood, the results are a reminder that common sample manipulations may affect the outcome of particle measurements. Method development should include evaluation of all sample handling steps. Comparison of NTA to DLS and aF4-MALS As NTA is a relatively new technique, it is important to compare it to other more established methods suitable for sub-micron particle analysis, such as DLS and aF4-MALS. The following aspects were considered when comparing the performance of the three analytical methods: size resolution, size accuracy, sensitivity, and dynamic range. A 2.5 mg/mL solution of stressed MAb1 was tested in parallel by NTA, DLS, and aF4-MALS. An overlay of the size distributions measured by the NTA and DLS techniques is presented in Figure 8. DLS was able to 436

aF4-MALS chromatogram of a 450 nm filtered MAb1 stressed sample. The later peak (20 –25 min) corresponds to elution of sub-micron particles. MALS analysis of the later peak shows a heterogeneous particle distribution in the 40-500 nm range (black open circles). The NTA generated distribution for the same sample (green curve) is presented for reference.

resolve two separate populations: the main peak (2–20 nm) that can be assigned to monomer/small oligomers of the MAb1 molecule, and a minor peak (100 – 6000 nm) that was assigned to protein particles present in solution. NTA was able to only detect the sub-micron particle population, and in that size range it shows a good size correlation with the reported DLS distributions. The analysis of the same sample by aF4-MALS showed two base-separated peaks, corresponding to populations of protein monomer and large aggregates. However, the light scattering data was negatively affected by signal from larger particles (data not shown); the dominant elution mode for these particles is steric, with larger particles exiting the channel first, contrary to the regular mode with smaller particles eluting faster) (28, 29). Therefore the side-by-side evaluation of aF4-MALS and NTA was done on a 450 nm filtered MAb1 stressed solution. The aF4-MALS results are presented in Figure 9; the main peak represents the elution of the protein monomer, while the later-eluting peak corresponds to the elution of sub-micron protein particles. Particle size was computed for a spherical shape to allow comparison with NTA or DLS, which rely on a similar approximation to generate size distribution information. The size range of the sub-micron particles detected by aF4-MALS covers a 40 –500 PDA Journal of Pharmaceutical Science and Technology


nm range. NTA measurements of the same filtered sample of stressed MAb1 are shown in green. Submicron particles detected by NTA cover the 50 –500 nm range, with most of the particles concentrated between 100 and 350 nm. The range of particles sizes reported by aF4-MALS is similar; however, the peak of its distribution is closer to 300 nm. It is worth pointing out that aF4-MALS does not directly compute particle concentrations, and data displayed in Figure 9 should not be interpreted as a true particle distribution but rather as a visualization tool for individual chromatogram slices. Attempts to gather DLS sub-micron particles size information for the filtered sample were unsuccessful. While regularization fitting of DLS data did show small amounts of particles in solution, size information about such species had low repeatability (data not shown). Additional DLS measurements of the filtered sample of stressed MAb1 were carried out with two other single cell instruments that confirmed the original observation: No reliable determination of submicron particles size distributions could be generated for the 450 nm filtered sample by employing DLS. The comparison of the three techniques discussed herein showed satisfactory correlations between the size distribution measurement results, but it also demonstrated limitations of the individual methods. When testing stressed MAb1 samples, the performance of the aF4-MALS was strongly affected by the presence of large (micron or larger) particles that elute sterically, along the entire chromatogram. On the other hand, when 450 nm filtered MAb1 samples were tested, DLS showed high variability in resolving the sub-micron particle populations. This shows that below a certain threshold of particle concentration, when the number fluctuations in the sampling volume start to affect the autocorrelation function, DLS performance becomes inadequate (30, 31). The NTA was able to perform good-quality measurements both in the presence of large particles and in cases when the concentration of sub-micron particles was relatively low. Additionally, it is worth noting that NTA provides sub-micron particle distributions as counts per volume directly, unlike both DLS and aF4-MALS.

the range of particle concentrations that can be optimally measured by NTA is 108 to 3 ⫻ 109 particles/mL. Size accuracy is excellent for particle standards smaller than 200 nm, but measurements of larger particles result in undersizing by as much as 20% from the nominal sizes. Based on our assessment, the phenomenon of photophoresis is at least partially responsible for this effect. We estimate that the effect of photophoresis will marginally interfere with measurements of sub-micron protein particles in protein solutions. Although NTA is acceptable for solutions of particle standards in water, particles detection and sizing in protein solutions are more challenging. Increased background signal, due to opalescence or increased protein concentration, may result in a loss of contrast and thus result in lower sensitivity and accuracy. Additionally, the performance of NTA appears to be molecule-specific. For some protein solutions, the range of protein concentrations that allows accurate NTA measurements appears to be limited to approximately 10 mg/mL. For higher protein concentration samples, dilution may be required to lower the background scattering and improve detection contrast (16). Nevertheless, dilution should be used with caution, given that it may perturb the original particle distributions. We have shown in this work that common sample handling operations, such as dilution or filtration, can affect the distributions of sub-micron particles in a molecule-specific manner. NTA performance is comparable with that of DLS and aF4-MALS. The key advantages of NTA are sensitivity, size resolution, ease of use, availability of information on individual particle counts, and a relatively broad dynamic range. Unlike NTA, the aF4-MALS measurements were affected by the presence of large (⬎1 micron) particles in solution. DLS showed lower sensitivity when measuring low abundance populations of sub-micron particles. Overall, NTA proves to be a valuable orthogonal technique for measurements of sub-micron particles in protein solutions and complements other analytical techniques used for the assessment of particle size distributions in protein solutions during biopharmaceutical development.

Conclusions

Acknowledgements

NTA was evaluated with suspensions of NIST-traceable polystyrene particle standards and MAb solutions of various concentrations. Based on particle standards data,

We thank Jared Bee, Richard Louis Remmele, Ken Miller, Ziping Wei and Mark Schenerman for their insightful comments while reviewing this manuscript.


437


Conflict of Interest Declaration The authors declare that they have no competing interests. References 1. Zolls, S.; Tantipolphan, R.; Wiggenhorn, M.; Winter, G.; Jiskoot, W.; Friess, W.; Hawe, A. Particles in therapeutic protein formulations, Part 1: overview of analytical methods. J. Pharm. Sci. 2012, 101, 914 –935. 2. Philo, J. S.; Arakawa, T. Mechanisms of protein aggregation. Curr. Pharm. Biotechnol. 2009, 10, 348 –351. 3. Mahler, H. C.; Friess, W.; Grauschopf, U.; Kiese, S. Protein aggregation: pathways, induction factors and analysis. J. Pharm. Sci. 2009, 98, 2909 –2934. 4. Kiese, S.; Papppenberger, A.; Friess, W.; Mahler, H. C. Shaken, not stirred: mechanical stress testing of an IgG1 antibody. J. Pharm. Sci. 2008, 97, 4347– 4366. 5. Wang, W.; Singh, S. K.; Li, N.; Toler, M. R.; King, K. R.; Nema, S. Immunogenicity of protein aggregates— concerns and realities. Int. J. Pharm. 2012, 431, 1–11. 6. Singh, S. K.; Afonina, N.; Awwad, M.; BechtoldPeters, K.; Blue, J. T.; Chou, D.; Cromwell, M.; Krause, H. J.; Mahler, H. C.; Meyer, B. K.; Narhi, L.; Nesta, D. P.; Spitznagel, T. An industry perspective on the monitoring of subvisible particles as a quality attribute for protein therapeutics. J. Pharm. Sci. 2010, 99, 3302–3321. 7. Rosenberg, A. S.; Verthelyi, D.; Cherney, B. W. Managing uncertainty: A perspective on risk pertaining to product quality attributes as they bear on immunogenicity of therapeutic proteins. J. Pharm. Sci. 2012, 101, 3560 –3567.

for subvisible and visible particles. Curr. Pharm. Biotechnol. 2009, 10, 373–381. 10. Li, Y.; Lubchenko, V.; Vekilov, P. G. The use of dynamic light scattering and brownian microscopy to characterize protein aggregation. Rev. Sci. Instrum. 2011, 82, 053106. 11. Helmbrecht, C.; Niessner, R.; Haisch, C. Photophoretic velocimetry—a new way for the in situ determination of particle size distribution and refractive index of hydrocolloids. Analyst 2011, 136, 1987–1994. 12. Saveyn, H.; De Baets, B.; Thas, O.; Hole, P.; Smith, J.; Van der Meeren, P. Accurate particle size distribution determination by nanoparticle tracking analysis based on 2-D Brownian dynamics simulation. J. Colloid Interface Sci. 2010, 352, 593– 600. 13. Filipe, V.; Hawe, A.; Jiskoot, W. Critical evaluation of Nanoparticle Tracking Analysis (NTA) by NanoSight for the measurement of nanoparticles and protein aggregates. Pharm. Res. 2010, 27, 796 – 810. 14. Huang, F.; Dempsey, C.; Chona, D.; Suh, J. Quantitative nanoparticle tracking: applications to nanomedicine. Nanomedicine. (Lond) 2011, 6, 693–700. 15. Bell, N. C.; Minelli, C.; Tompkins, J.; Stevens, M. M.; Shard, A. G. Emerging techniques for sub-micron particle sizing applied to Stober silica. Langmuir 2012, 28 (29), 10860 –10872. 16. Patois, E, Capelle M.A.H, Palais C., Gurney, R., and Arvinte, T. Evaluation of nanoparticle tracking analysis (NTA). J.Drug.Del.Sci.Tech., 2012, 22, 427– 433.

8. Philo, J. S. A critical review of methods for size characterization of non-particulate protein aggregates. Curr. Pharm. Biotechnol. 2009, 10, 359 –372.

17. Kumru, O. S.; Liu, J.; Ji, J. A.; Cheng, W.; Wang, Y. J.; Wang, T.; Joshi, S. B.; Midduagh, C. R.; Volkin, D. B. Compatibility, physical stability, and characterization of an IgG4 monoclonal antibody after dilution into different intravenous administration bags. J. Pharm. Sci. 2012, 101, 3636 –3650.

9. Narhi, L. O.; Jiang, Y.; Cao, S.; Benedek, K.; Shnek, D. A critical review of analytical methods

18. Jachimska, B.; Wasilewska, M.; Adamczyk, Z. Characterization of globular protein solutions by

438

PDA Journal of Pharmaceutical Science and Technology


dynamic light scattering, electrophoretic mobility, and viscosity measurements. Langmuir 2008, 24, 6866 – 6872. 19. He, F.; Becker, G. W.; Litowski, J. R.; Narhi, L. O.; Brems, D. N.; Razinkov, V. I. Highthroughput dynamic light scattering method for measuring viscosity of concentrated protein solutions. Anal. Biochem. 2010, 399, 141–143. 20. Wei, Z.; McEvoy, M.; Razinkov, V.; Polozova, A.; Li, E.; Casas-Finet, J.; Tous, G. I.; Balu, P.; Pan, A. A.; Mehta, H.; Schenerman, M. A. Biophysical characterization of influenza virus subpopulations using field flow fractionation and multiangle light scattering: correlation of particle counts, size distribution and infectivity. J. Virol. Methods 2007, 144, 122–132. 21. Qureshi, R. N.; Kok, W. T. Application of flow field-flow fractionation for the characterization of macromolecules of biological interest: a review. Anal. Bioanal. Chem. 2010, 399-4, 1401–1411. 22. Liu, J.; Andya, J. D.; Shire, S. J. A critical review of analytical ultracentrifugation and field flow fractionation methods for measuring protein aggregation. AAPS. J. 2006, 8, E580 –E589. 23. Zhao, B. S.; Koo, Y. M.; Chung, D. S. Separations based on the mechanical forces of light. Anal. Chim. Acta 2006, 556, 97–103. 24. Watarai, H.; Monjushiro, H.; Tsukahara, S.; Suwa, M.; Iiguni, Y. Migration analysis of micro-particles in liquids using microscopically designed external fields. Anal. Sci 2004, 20, 423– 434.


25. Zolls, S.; Gregoritza, M.; Tantipolphan, R.; Wiggenhorn, M.; Winter, G.; Friess, W.; Hawe, A. How subvisible particles become invisible-relevance of the refractive index for protein particle analysis. J. Pharm. Sci 2013, 102, 1434 –1446. 26. Salinas, B. A.; Sathish, H. A.; Bishop, S. M.; Harn, N.; Carpenter, J. F.; Randolph, T. W. Understanding and modulating opalescence and viscosity in a monoclonal antibody formulation. J. Pharm. Sci 2010, 99, 82–93. 27. Demeule, B.; Messick, S.; Shire, S. J.; Liu, J. Characterization of particles in protein solutions: reaching the limits of current technologies. AAPS. J. 2010, 12, 708 –715. 28. Williams, P. S.; Giddings, J. C. Theory of fieldprogrammed field-flow fractionation with corrections for steric effects. Anal. Chem. 1994, 66, 4215– 4228. 29. S.Kim Ratanathanawongs and Williams Dean Lee. Field-flow fractionation of proteins, polysaccharides, synthetic polymers, and supramolecular assemblies. J.Sep.Sci., 2006, 29, 1720 –1732. 30. Bloomfield, V. A.; Lim, T. K. Quasi-elastic laser light scattering. Methods Enzymol. 1978, 48, 415– 494. 31. Schersch, K.; Betz, O.; Garidel, P.; Muehlau, S.; Bassarab, S.; Winter, G. Systematic investigation of the effect of lyophilizate collapse on pharmaceutically relevant proteins I: stability after freeze-drying. J. Pharm. Sci 2010, 99, 2256 –2278.

439


An Authorized User of the electronic PDA Journal of Pharmaceutical Science and Technology (the PDA Journal) is a PDA Member in good standing. Authorized Users are permitted to do the following: ·Search and view the content of the PDA Journal ·Download a single article for the individual use of an Authorized User ·Assemble and distribute links that point to the PDA Journal ·Print individual articles from the PDA Journal for the individual use of an Authorized User ·Make a reasonable number of photocopies of a printed article for the individual use of an Authorized User or for the use by or distribution to other Authorized Users Authorized Users are not permitted to do the following: ·Except as mentioned above, allow anyone other than an Authorized User to use or access the PDA Journal · Display or otherwise make any information from the PDA Journal available to anyone other than an Authorized User ·Post articles from the PDA Journal on Web sites, either available on the Internet or an Intranet, or in any form of online publications ·Transmit electronically, via e-mail or any other file transfer protocols, any portion of the PDA Journal ·Create a searchable archive of any portion of the PDA Journal ·Use robots or intelligent agents to access, search and/or systematically download any portion of the PDA Journal ·Sell, re-sell, rent, lease, license, sublicense, assign or otherwise transfer the use of the PDA Journal or its content ·Use or copy the PDA Journal for document delivery, fee-for-service use, or bulk reproduction or distribution of materials in any form, or any substantially similar commercial purpose ·Alter, modify, repackage or adapt any portion of the PDA Journal ·Make any edits or derivative works with respect to any portion of the PDA Journal including any text or graphics ·Delete or remove in any form or format, including on a printed article or photocopy, any copyright information or notice contained in the PDA Journal

Characterization of submicron (0.1-1 μm) particles in therapeutic proteins by nanoparticle tracking analysis.

Characterization of Nanoparticle Tracking Analysis for Quantification and Sizing of Submicron Particles of Therapeutic Proteins.

Dark field nanoparticle tracking analysis for size characterization of plasmonic and non-plasmonic particles.

Evaluation of Nanoparticle Tracking for Characterization of Fibrillar Protein Aggregates.

Minimizing biases associated with tracking analysis of submicron particles in heterogeneous biological fluids.

Protein-polyelectrolyte interactions: Monitoring particle formation and growth by nanoparticle tracking analysis and flow imaging microscopy.

Mass isotopomer analysis: theoretical and practical considerations.

heating module for tissue chambers and solutions: theoretical considerations and practical design.

Practical considerations for volumetric wear analysis of explanted hip arthroplasties.

Analysis of nanoparticle-protein coronas formed in vitro between nanosized welding particles and nasal lavage proteins.

Broad spectrum analysis of polar and apolar organic compounds in submicron atmospheric particles.

Quantitative tracking of protein trafficking to the nucleus using cytosolic protein delivery by nanoparticle-stabilized nanocapsules.

Practical considerations for maintaining germplasm in maize.

Precipitation of submicron charged particles in human lung airways.

Acoustic separation of submicron solid particles in air.

Measuring protein dynamics in live cells: protocols and practical considerations for fluorescence fluctuation microscopy.

Characterization of extracellular vesicles in plasma of pregnant women using multicolor flow cytometry and nanoparticle tracking analysis.

Size and concentration analyses of extracellular vesicles by nanoparticle tracking analysis: a variation study.

Practical Considerations for Implementing Research Recruitment Etiquette.

Acupuncture: practical considerations.

Practical considerations for ACR MRI accreditation.

Acoustic detection, tracking, and characterization of three tornadoes.

Analysis of Android Device-Based Solutions for Fall Detection.

Food safety considerations for innovative nutrition solutions.