Abstract
Particles isolated from a pre-filled syringe containing a protein-based solution were identified as aggregated protein and tungsten. The origin of the tungsten was traced to the tungsten pins used in the supplier's syringe barrel forming process. A tungsten recovery study showed that the vacuum stopper placement process has a significant impact on the total amount of tungsten in solutions. The air gap formed in the syringe funnel area (rich in residual tungsten) becomes accessible to solutions when the vacuum is pulled. Leachable tungsten deposits that were not removed by the supplier's wash process are concentrated in this small area. Extraction procedures used to measure residual tungsten in empty syringes would under-report the tungsten quantity unless the funnel area is wetted during the extraction. Improved syringe barrel forming and washing processes at the supplier have lowered the residual tungsten content and significantly reduced the risk of protein aggregate formation. This experience demonstrates that packaging component manufacturing processes, which are outside the direct control of drug manufacturers, can have an impact on the drug product quality. Thus close technical communication with suppliers of product contact components plays an important role in making a successful biotherapeutic.
1. Introduction
In recent years the biopharmaceutical landscape has changed dramatically, migrating from a few products with relatively low protein concentration, which were based on naturally occurring proteins, to a large number of products of all different protein types, concentrations, and formulations (1, 2). Protein aggregation is encountered at almost every stage of the protein drug product manufacturing process, including fermentation, purification, formulation, and fill and finish (3). There are many environmental factors, such as temperature, agitation, and pH, that can trigger protein aggregation (4). In addition, during these biotherapeutic processing steps, protein pharmaceutics often come in contact with a variety of surfaces or materials, which can also induce or facilitate protein aggregation.
As the types of therapeutic proteins have evolved, the preferred container–closure systems also have been moving from vials to pre-filled syringes (5, 6). Changes in the complexity of the products and the containers in which they are stored have introduced greater complexity in the interactions of the protein with the packaging components and consequences of these interactions on product stability. For example, with a pre-filled syringe, the syringe barrel is often made of borosilicate glass, silicone oil is applied to the barrel for plunger lubrication, the plunger stopper is made of elastomer and may be coated with fluoropolymer and silicone, the syringe needle is made of stainless steel, UV-cured adhesive is used to fix the needle to the syringe barrel, and the needle shield is made of yet another elastomer (7).
We have recently found that another material, tungsten at trace levels, is also present in the syringe barrels, making the environment of the pre-filled syringe even more challenging. Tungsten pins are used in the manufacturing of glass pre-filled syringe barrels to form the small diameter fluid path at the barrel tip. We show here that residual tungsten compounds from the syringe tip forming process are associated with protein aggregation and particles. In this paper, we study the impact of the syringe barrel manufacturing and product fill and finish processes on product tungsten exposure, and we demonstrate that changes in the syringe barrel forming and washing processes can significantly decrease product exposure to tungsten.
2. Materials and Methods
Tungsten Identification in Protein Particles
Pre-filled syringes containing protein-based solutions (an alpha helical protein at buffer solution pH ∼4.0) with visible particles were selected for the analysis as follows. The protein solution was filled into the pre-filled syringes (glass syringe with staked needles) in a clean room environment (class 100) and kept at 2–8 °C. The pre-filled syringes were then visually inspected for particulates (the time between pre-filled syringe preparations to visual inspection was usually greater than 24 h); syringes containing particles were selected for further analysis.
Protein solution from the selected pre-filled syringes was filtered with a Millipore nitrocellulose membrane filter (0.45 μm, Millipore AABG02500) in a clean room environment (Class 100). The plunger of the pre-filled syringe was removed and the liquid contents were poured onto the filter and drained through the filter. The micro-particles were retained on the filter. The filter was rinsed with de-ionized water to wash away any remaining protein solution and dried in a laminar flow hood.
A Carl Zeiss stereomicroscope with magnification of 56× was employed to examine the morphology and size of the micro-particles retained on the membrane filter.
Infrared (IR) spectra of the micro-particles were obtained with a Continμmm IR microscope attached on a Magna 550 IR optical bench (Thermal Nicolet, Madison, WI). The micro-particles were transferred under the stereo-microscope onto a 13 mm × 2 mm potassium bromide (KBr) disk and then sandwiched between two KBr disks in a SpectraTech micro-compression cell for IR spectral acquisition. The spectral resolution of the IR spectrum is 4 cm−1. Each IR spectrum represents the signal averaged from 128 scans. The reference IR spectrum of the protein product in aqueous solution was also obtained with the same procedure. The protein solution was dried on a glass slide and transferred onto the KBr pellet.
Scanning electron microscopy (SEM) images of the micro-particles were obtained with a Zeiss EVO 40 variable pressure SEM. The accelerating voltage was set at 20 kV. The chamber pressure was held at approximately 50 Pa in order to minimize the charging effect that occurs on non-conductive materials. The SEM images of the micro-particles were obtained with a variable pressure secondary electron (VPSE) detector, which enables true secondary imaging under variable pressure conditions. In addition to generating ions during collision, the secondary electron collisions also produce photons, which are collected by the VPSE detector to produce a true secondary electron image (8).
The identification of elements in the samples was accomplished with the attached energy dispersive spectroscope (EDS), an Oxford INCA Energy 300 attached to the Zeiss SEM. During the SEM imaging, the focused incident electron beam simultaneously interacts with the tightly bound inner shell electron of the samples, ejecting the atomic electron of the inner shell and leaving a vacancy in that shell. The transition of the electron from one shell to the vacant shell is accompanied by the emission of characteristic X-rays, which have specific energies and wavelengths and are characteristic of the particular element (8). The EDS spectrum therefore identifies the elemental composition of the material.
The operating software of the EDS has a Cameo© feature, which generates a map of the distribution of elements in a sample, with the compositional variations in a sample visualized by offsetting the X-ray spectrum into the visible range and assigning different colors to the different elements. This feature allows one to see the chemical composition and topography in one high-resolution image. The fine detail of the electron image is shown together with a full color overlay, which can reveal subtle composition differences and distributions.
Tungsten Recovery from the Syringe Barrel
To demonstrate the effect of vacuum stopper placement on product exposure to the tungsten-rich funnel area, aqueous dye solution was prepared by dissolving rhodamine B in de-ionized water at ∼1% (w/v). The base of the syringe needle cover was cut off and removed to expose the “funnel” of the syringe area that normally is obscured. Approximately 0.3 mL of dye solution was first filled into the syringe barrel. After filling, a plunger was inserted using a long, disposable hypodermic needle to vent the air. A photomicrograph of the “as-filled” funnel was taken using a Zeiss stereomicroscope with attached digital camera. Vacuum plunger placement was then simulated by withdrawing the plunger to the open end of the barrel while maintaining the syringe in a needle-down orientation. This reduced the pressure inside the headspace to approximately 0.1 bar. The plunger then was released and allowed to return to its equilibrium position. A second image of the funnel was taken after the simulated vacuum stopper placement had been completed.
Based on this demonstration, a study was designed to compare extraction of tungsten with and without vacuum stopper placement using protein buffer or de-ionized water as the extraction medium. 0.5 mL protein buffer (solution pH ∼4.0 containing surfactant) or de-ionized water was filled into the syringe barrels using an Eppendorf Repeator hand filler. After the fill was completed, plungers were placed into half of the filled syringes using an Autoclavable Stopper Placement Unit (ASPU) (Impro Systems, Inc.) at a vacuum level of ∼20 mm Hg. The rest of the syringe barrels were sealed with Parafilm to prevent contamination and evaporation. All syringes were then moved to a 2–8 °C cold room for incubation for 96 h before analyzing tungsten concentrations.
A Perkin Elmer/SCIEX Elan DRC II Inductively Coupled Plasma Mass Spectrometry (ICP/MS) equipped with an autosampler (AS-93Plus) was used in this work. The instrument was optimized following the standard procedures recommended by the vendor. Sampling was performed at sample flush 45 s/24 rpm, read delay 25 s/20 rpm, and wash 90 s/35 rpm with 0.5% nitric acid. Three replicate measurements per sample were acquired at 60 sweeps/reading. Tungsten standard solutions (Cat 3566397-100) and iridium standard solutions (Cat 207209-100) were purchased from Sigma-Aldrich. Nitric acid Optima (Cat A467-500) was purchased from Fisher Scientific. Water from a laboratory water purification system (Millipore Milli-Q, ≥18.0 MΩcm) was used. Standard calibration solutions of tungsten were prepared at concentrations of 0.5, 1.0, 5.0, 10, 25.0, 50.0, 100.0, and 200.0 ppb in 0.5% nitric acid solution. Iridium was used as an internal standard at 50 ppb. Analysis monitored masses of 184 and 186 for tungsten and 193 m/z for iridium. Calibration curves had correlations (R2) ≥ 0.9999 for tungsten 184. Samples were diluted to 10 mL with 0.5% nitric acid and spiked to contain 50 ppb iridium. A blank 0.5% nitric acid solution was measured prior to each sample analysis to check background noise. The method's accuracy was tested against four tungsten standard samples at concentrations of 0.8, 4, 15, and 75 ppb tungsten. The method was accurate to ±7% of the theoretical concentration of tungsten for all the tungsten standard levels except for the 0.8 ppb level (±18%).
3. Results and Discussion
Pre-filled syringes of a protein solution exhibited particles. The particles were analyzed after filtration onto membranes and the nature of the materials found led to further studies relating to components of the pre-filled syringes in which they were originally contained.
Particle Identification
Figure 1 shows the optical micrograph of the particles obtained from the syringes. The particle size ranges from 10 to 100 μm with irregular shape. Figure 2 displays the IR spectrum of one representative micro-particle, along with a reference protein IR spectrum of the protein product contained in the pre-filled syringe for comparison. The IR spectrum of the representative micro-particle is consistent with it containing protein, as indicated by the prominent NH stretching (3300 cm−1), Amide I (1650 cm−1), Amide II (1540 cm−1), and Amide III (1245 cm−1) signals. It matches well with the protein IR spectrum except for a few small peaks in the 1000 to 800 cm−1 region, which could not be assigned to the protein molecule. These IR spectroscopic features suggest that the micro-particle is a mixture of the protein product with some other materials.
Optical micrograph of micro-particles from the pre-filled syringe retained on the membrane filter.
The FTIR spectrum of a micro-particle (top) and a reference FTIR spectrum of the protein product (bottom).
Figure 3 shows the SEM micrograph of another representative micro-particle and its EDS spectrum. The EDS spectrum of the particle unambiguously identifies the presence of tungsten, which exhibits the characteristic Mα and Lα peaks centered at 1.775 KeV and 8.398 KeV, respectively (8). It is also noted that there are weak signals from sulfur and nitrogen elements as well that likely to come from the protein.
The SEM micrograph of (A) one of the micro-particles and (B) its EDS spectrum.
Figure 4 shows the Cameo© map image of one of the particles. It shows an even distribution of the tungsten throughout the micro-particle. In Cameo© images, areas where tungsten is present are mapped to a distinctive blue color. In this case, the entire particle is uniformly blue. This indicates that the tungsten is not concentrated in one spot, but is distributed evenly throughout the protein particle. The finding of the tungsten in the particle explains the three small peaks at 954, 906, and 808 cm−1 respectively, in the IR spectrum; they can be assigned to the tungsten ions (9).
The Cameo© Map image shows the even distribution of the tungsten fine particles throughout the protein particle. Figure 4A is only the Cameo© Map of tungsten fine particles, which is represented by the blue color. Figure 4B is the Cameo© Map of tungsten particles overlaid with the SEM images of the protein particle for clarity. Figure 4C is the EDS spectrum of the protein particle with tungsten fine particles embedded.
The protein solutions containing particles were also analyzed by other analytical techniques including size exclusion high-performance liquid chromatography, light scattering, and gel electrophoresis (data not shown), and soluble protein species were confirmed to be present in the solutions. The protein analytical data are not within the scope of this paper, and they are included in the mechanism study paper by Jiang et al. (10).
Identification of the Source of Tungsten Contained in Particles
The presence of tungsten as the only identified foreign material in the protein particles suggested that tungsten may have caused the protein particle formation. Consultation with the pre-filled syringe manufacturer confirmed that tungsten pins are used in the forming process of the syringe barrels. At glass-forming temperatures, in the order of 1000 °C, tungsten pins form various soluble and insoluble species that can react with vaporized oxides from the hot glass (personal communication). These tungsten-rich compounds can deposit on the interior of the syringe, primarily in the fluid path near the tip of the barrel.
During the manufacturing process for ready-to-fill glass syringe barrels, the barrels are rinsed with jets of 80 °C water for injection prior to final packaging and sterilization. However, the tungsten deposits may not be sufficiently soluble to be fully removed by the washing process. In addition, the amount and solubility of the tungsten residue may vary from one forming machine to another and from lot to lot of syringes depending on flame settings or other forming process parameters. The possibility that significant partially soluble tungsten residues could be present in the syringe after washing was confirmed by the analysis of white deposits seen in the funnel area of empty syringe barrels. Figure 5 shows a micrograph of the neck of a used product syringe barrel from a syringe that contained protein particles. The deposits contain a small, white residue (∼30 μm particle size) inside the glass funnel region. The subsequent SEM/EDS analysis confirmed that the white deposits clearly contain tungsten. Similar white deposits were also observed in unused syringe barrels (pictures not shown).
The micrograph of the syringe neck section and a residue found inside of the syringe neck (5A). (See circles). The SEM backscattering image of the residue (5B) and its EDS spectrum (5C).
The ability of the tungsten species produced by the manufacturing process to induce aggregation of protein was verified by co-incubation of protein with tungsten pin extract; studies are described in the paper by Jiang et al. (10).
In the current case, the protein solution had been stored in a variety of lots of syringes previously without aggregation or particle formation, and yet it was only recently that the particles appeared. This new observation of protein particulates suggests something has changed to cause the sudden increase of protein aggregates (the proteinaceous particulates). Through a thorough review of syringe manufacturing history and records, it was found that the supplier had made several ostensibly minor changes in the syringe barrel manufacturing process that apparently resulted in a shift to a higher level of residual tungsten without the supplier's knowledge at the time (personal communication). Through the investigative efforts of the syringe barrel supplier, it was discovered that changes in tungsten pin supplier, the syringe barrel forming process parameters, and washing process parameters together resulted in the increased level of residual tungsten in the empty syringes (personal communication).
Tungsten Extraction Procedures To Assess Levels of Residual Tungsten in Pre-filled Syringes
It is essential that an appropriate procedure be developed to provide an accurate and reproducible assessment of residual tungsten levels in empty syringes to be used with tungsten-sensitive proteins.
We investigated the impact of the syringe filling process on the total amount of residual tungsten extracted from the syringe barrel. The syringe filling and stopper placement process is illustrated in Figure 6. After filling solution into the empty syringe barrel (shown in Step A of Figure 6), a vacuum was pulled in the chamber of the stopper placement unit; both the inside and outside of the syringe barrel were under vacuum at this step. The stopper was then partially inserted into the top of the syringe barrel via the insertion pins (not shown) of the stopper placement unit (Step B). In Step C, the chamber was vented and the space outside the syringes was exposed to atmospheric pressure. Due to the pressure difference inside and outside of the syringe barrel, the stopper was pulled down to its final position in the syringe.
Simulated syringe filling and stoppering process.
Because the tungsten pin is primarily in contact with the barrel funnel area (to form the needle hole) during the barrel forming process, the barrel funnel area could potentially be “tungsten rich”. We have specifically investigated if the vacuum plunger placement process has any impact on the tungsten content of solution exposed to the funnel area. A dye solution was used to simulate the behavior of protein solution during the stopper placement process for the pre-filled syringe. Figure 7 compares the different pattern of solution contact with the barrel funnel area prior to (A) and after (B) simulated stopper placement. When the aqueous dye solution is filled into the syringe barrel, there is an air gap trapped in the “funnel area” of the barrel as shown in Figure 7 (A). When the vacuum is pulled, the trapped air expands sufficiently that most of the volume rises through the liquid as a bubble and escapes into the vacuum chamber. When the stopper is in its final position, the protein solution is in contact with the “funnel” area as shown in Figure 7 (B). Figure 7 indicates the vacuum-pulling process may increase solution exposure to the funnel area, which subsequently could lead to leaching of more tungsten into the solution.
Dye simulation shows the vacuum pulled during stopper placement process induces solution contact with the syringe tip area. (A) right after syringe fill; (B) after vacuum stopper placement. Note: Portion of needle shield removed to expose “funnel” area.
To further substantiate the impact of the vacuum-pulling process on residual tungsten recovery, tungsten extraction experiments were conducted comparing different solutions (water vs protein buffer), and vacuum stopper placement on the tungsten recovery. One hundred syringe barrels were filled with water or protein buffer (solution pH ∼ 4.0 containing surfactant). Half of the syringes for each solution were subjected to the vacuum stopper placement process. The other half was analyzed without vacuum exposure. The tungsten concentration was analyzed on all syringes after incubation at 2–8 °C for ∼96 h; the tungsten concentration results from 45 samples for each condition are shown in Figure 8. When water was filled into the pre-filled syringes, the average tungsten concentration after storage at 2–8 °C was 241 ppb (standard deviation 203 ppb) without pulling a vacuum, and the average was 356 ppb (standard deviation 233 ppb) with vacuum. When product buffer was filled into the pre-filled syringe, the average tungsten concentration after storage at 2–8 °C was 163 ppb (standard deviation 173 ppb) without pulling a vacuum, and the average was 279 ppb (standard deviation 288 ppb) with vacuum.
Tungsten recovery comparison with water, product buffer (placebo), and vacuum: (◆) using water without pulling vacuum; (■) using water with pulling vacuum; (▴) using product buffer without pulling vacuum; (●) using product buffer with pulling vacuum.
A statistical pair-wise comparison was performed on the tungsten results and the statistical P-value was calculated for a pair-wise comparison between all experimental conditions; the results are listed in Table I. The P-value needs to be <0.05 to declare statistical significance (11). Results in Table I showed that a significant difference exists for the quantity of tungsten extracted into solution when vacuum stopper placement is used compared to samples not exposed to the vacuum process. However, there is no significant difference in the amount of tungsten extracted into the protein buffer when compared to water.
Statistical P-value for Pair-wise Comparison between Conditions
Vacuum stopper placement draws the solution into the funnel area of the barrel tip, as illustrated in Figure 7. The volume contained in this area is on the order of 5 μL, which is only 1% of the total protein solution volume. These results demonstrate that the leachable tungsten deposits that were not removed by the supplier's wash process are concentrated in this small area. As such, an extraction procedure used to measure residual tungsten in empty syringes would under-report the tungsten quantity unless the test samples were exposed to vacuum after filling, as in the protocol described above, to ensure that the funnel area is wetted during the extraction. The results also show that water may be used as the extraction medium for such a procedure.
Based on the finding that the vacuum process increases residual tungsten recovery, the syringe barrel manufacturer has implemented an analytical method for residual tungsten in the supplied barrels. This method has been used by the supplier to validate improved forming and washing processes resulting in substantial reduction in residual tungsten (the detailed method from the supplier is proprietary information, and is outside the scope of this paper). The method is also the basis for an agreed-upon specification for residual tungsten as a barrel lot release criterion.
4. Conclusions
We report here the observation and investigation of tungsten-induced protein particle formation using pre-filled syringes as the container system/delivery device. During the glass barrel manufacturing process, tungsten pins are used to form the fluid path in the barrel tip. This resulted in the deposition of various tungsten species (the current study did not distinguish between soluble and insoluble forms of tungsten) on the syringe barrel, especially in the funnel region. Our results demonstrate that the leachable tungsten deposits are concentrated in this small area. Vacuum stopper placement during the fill and finish process facilitated the contact of the protein solution with the tungsten residue not removed by the supplier's wash process and is the basis for a standardized, accurate, and reproducible tungsten extraction procedure. Based on this information, the syringe barrel supplier has implemented improved barrel forming and washing processes that have significantly reduced residual tungsten.
These studies demonstrate the impact that container components can have on the quality of the final drug dosage form. In this case an increase in the amount of particles was tied to unintended consequences of “minor” changes in the syringe barrel manufacturing process. This highlights the necessity for clear and open communication between therapeutic drug manufacturers and their packaging component suppliers regarding process changes. The adage “the process is the product” is as valid for container systems as it is for the formulated drug product.
Acknowledgements
The work presented in this paper was funded by and conducted in Amgen Inc. The authors gratefully acknowledge Frank Ye and Jerry Sparks for their input to the experiment design. The authors would also like to thank David Brems, Joseph Phillips, Narinder Singh, Steve Ruhl, and Vinay Radhakrishnan for their review and discussion of this manuscript. The authors acknowledge the extensive cooperation of the syringe supplier in the investigation and resolution of this issue.
- © PDA, Inc. 2010
Jump to section
Related Articles
Cited By...
- Holistic Extractables and Leachables Program: Evaluations of Prefilled Syringe Systems for Biotechnology Products
- A Comparison of Protein Stability in Prefillable Syringes Made of Glass and Plastic
- Characterization of Protein Aggregating Tungstates: Electrospray Mass Spectrometry Analysis of Extracts from Prefilled Syringes and from Tungsten Pins Used in the Manufacture of Syringes
- Development and Justification of a Risk Evaluation Matrix To Guide Chemical Testing Necessary To Select and Qualify Plastic Components Used in Production Systems for Pharmaceutical Products
- Creating a Holistic Extractables and Leachables (E&L) Program for Biotechnology Products
- Materials in Manufacturing and Packaging Systems as Sources of Elemental Impurities in Packaged Drug Products: A Literature Review
- Development of an Inductively Coupled Plasma Mass Spectrometry Method for Quantification of Extracted Tungsten from Glass Prefilled Syringes Used as a Primary Packaging for Pharmaceutical and Therapeutic Protein Products
- Inhibition of Tungsten-Induced Protein Aggregation by Cetyl Trimethyl Ammonium Bromide