Abstract
Container closure integrity (CCI) is a critical factor to ensure that product sterility is maintained over its entire shelf life. Assuring the CCI during container closure (C/C) system qualification, routine manufacturing and stability is important. FDA guidance also encourages industry to develop a CCI physical testing method in lieu of sterility testing in a stability program. A mass extraction system has been developed to check CCI for a variety of container closure systems such as vials, syringes, and cartridges. Various types of defects (e.g., glass micropipette, laser drill, wire) were created and used to demonstrate a detection limit. Leakage, detected as mass flow in this study, changes as a function of defect length and diameter. Therefore, the morphology of defects has been examined in detail with fluid theories. This study demonstrated that a mass extraction system was able to distinguish between intact samples and samples with 2 μm defects reliably when the defect was exposed to air, water, placebo, or drug product (3 mg/mL concentration) solution. Also, it has been verified that the method was robust, and capable of determining the acceptance limit using 3σ for syringes and 6σ for vials.
LAY ABSTRACT: Sterile products must maintain their sterility over their entire shelf life. Container closure systems such as those found in syringes and vials provide a seal between rubber and glass containers. This seal must be ensured to maintain product sterility. A mass extraction system has been developed to check container closure integrity for a variety of container closure systems such as vials, syringes, and cartridges. In order to demonstrate the method's capability, various types of defects (e.g., glass micropipette, laser drill, wire) were created in syringes and vials and were tested. This study demonstrated that a mass extraction system was able to distinguish between intact samples and samples with 2 μm defects reliably when the defect was exposed to air, water, placebo, or drug product (3 mg/mL concentration) solution. Also, it was verified that the method showed consistent results, and was able to determine the acceptance limit using 3σ for syringes and 6σ for vials.
- Cartridge
- Container closure integrity
- Container closure system
- Defect
- Fluid theory
- Glass
- Leak
- Mass extraction
- Microorganism
- Packaging
- Parenteral
- Pharmaceutical
- Protein
- Sterility
- Syringe
- Vacuum, Vial
Introduction
Container closure integrity (CCI) is a critical attribute of sterile products. If a C/C system has defects, channels can be created in the C/C system and microorganisms can ingress through the channels, affecting product safety. Therefore, manufacturing processes for sterile products must ensure that appropriate controls are in place to ensure sterility, including maintenance of CCI. In order to ensure CCI, C/C systems should be qualified and sealing processes should be validated. Furthermore, CCI of finished products prepared from qualified C/C systems and sealing processes should be tested periodically throughout the product's shelf life. Regulatory requirements including those of the U.S. Food and Drug Administration (FDA) require performance of periodic process simulations (commonly referred to as “media fills”) as well as batch-specific product sterility testing to confirm product sterility (1). In addition, validation studies in support of the microbial integrity of the C/C system are required since product sterility testing alone is not sufficient to ensure that products maintain their sterility throughout the manufacturing and dating period (2, 3). The U.S. Pharmacopeia (USP) and Parenteral Drug Association (PDA) also recommend that CCI physical testing should be performed at three phases of the product lifecycle: initial C/C system qualification, routine manufacturing, and shelf life (4, 5). Testing methods to check the CCI of products from manufacture and products stored stably should be developed to assure sterility of products and to meet regulatory expectations. Physical CCI testing methods can be used to meet the testing requirements for all phases of the product's lifecycle.
One of the major questions of CCI physical testing method development is what detection limit (or sensitivity) the method should achieve. The smallest diameter of microorganisms ranges from 0.4 to 0.7 μm (e.g., Pseudomonas aeruginosa, Escherichia coli, and Brevundimonas diminuta) (6). Kirsch et al. reported the relationship between microbial ingress and defect size (7). Kirsch showed that microorganisms can ingress through 0.4 and 0.7 μm diameter glass micropipettes with a detection probability of approximately 11% and 64%, respectively. Burrell et al. showed that no microorganism ingress occurs through 2–5 μm diameter microtubes but the microorganisms can ingress through microtubes of 10 μm and larger (8). Morrical et al. reported no microorganism ingress through micro-hole plates up to 2 μm but 20% and 90% detection can be achieved from 4 and 15 μm micro-hole plates, respectively. They studied another type of simulated defect, wire in rubber stopper/vial/aluminum crimp seal. They reported microorganism can ingress through channels created from 20 and 60 μm outer diameter wire in the seal with a detection probability of 35% and 100%, respectively (9). From these studies, it was ascertained that the probability of microorganism ingress can be significantly different depending on the shape of the defect and the testing method employed. Like microbial ingress testing methods, other CCI physical testing methods demonstrate varying results from different shapes of defects. Therefore, it is important that defects be defined when demonstrating the method's capability.
Ideally, the CCI method may need to detect any defects that can allow microbial ingress, which is around 0.4 μm. However, there is no practical testing technology currently available that can achieve this sensitivity reliably within a short testing time and a non-destructive test method as desired. In practice, typical defects for C/C systems do not result in single channels of defined diameter (e.g., 0.4–2 μm) but rather manifest themselves as tortuous paths of varying diameters due to such things as glass cracks, insufficient compression/sealing, or other component-related issues. Even though it is incumbent on the manufacturer to minimize the occurrence of and to remove such defects in the inspection process, the overall true risk that they pose to product sterility is arguable. Considering physical CCI test method limitations and the practical relationship between a quantitative limit of detection and the nature of defects expected in a manufacturing setting, a CCI method demonstrating a detection limit of 2–5 μm, with 100% detection probability, may be sufficient to ensure CCI for sterile products. This criterion may be supported by an understanding that a physical test method such as mass extraction is in effect detecting the overall/cumulative leak area in the container as opposed to a single defined leak of known surface area. In this paper, a new CCI physical testing method, a mass flow measurement technology known as mass extraction is introduced as a limit test, and International Conference on Harmonization (ICH) method validation characteristics, robustness, and detection limit are determined using a mass extraction measurement system.
A scientific validation study should be performed to demonstrate the detection limit of any CCI physical testing method. Leakage detection as well as the propensity for microbial ingress can be affected by such factors as geometry (e.g., diameter and length), properties of contents (e.g., air and liquid), and pressure difference. Therefore, simulated defects should be understood well to demonstrate the detection limit of CCI physical testing methods. Several simulated defects should be examined in detail with suggested fluid transport analytical models. When the same defect is located in different container closure systems (e.g., syringes, vials or cartridges) but everything else is the same, the same response of mass flow in this study will be generated from the defect. However, background noise (or baseline) will be slightly different between different C/C systems.
Mass Extraction Measurement Technologies
A mass extraction instrument measures the mass flow extracted from a sample under test. The micro-flow sensor is designed to detect a flow in transitional and molecular flow regimes (10⇓–12) and it is calibrated using primary NIST traceable standards. The flow is measured in mass flow units of μg/min. The instrument can be verified with a simple suitability check, with an empty chamber and 2 μm glass micropipette.
Principle
The mass extraction measurement is based on the mass conservation law, where the sensor measures the flow rate of gas extracted from a sample in the testing chamber under a vacuum. The mass flow rate at steady state conditions extracted from the test chamber is equal to the container (sample) leakage into the test chamber assuming outgassing from the sample and that system external leakage are not significant to the test.
The micro-flow sensor has a cross-section as shown in Figure 1, where the gas molecules are forced into an annular capillary gap with the width h. The gas transport through this capillary gap (h) results in momentum loss. This momentum loss is proportional to the gas flow rates and is dependent on the flow regime. This momentum loss generates an electrical signal which is calibrated against independent primary flow standards (NIST traceable, per the requirements of ISO 17025). The calibration coefficients are stored within the sensor and hence the sensor can display the true flow rates of gas flowing through the sensor during normal operation.
Cross-section of the micro-flow sensor element.
The sensor measurement is dependent on the flow regime in which it is operating. The different flow regimes can be characterized by Knudsen Number (Kn) as:
where
λ h = the annulus capillary width (μm) of the micro-flow sensor capillary flow path
λ = mean free-flow path (μm) of the gas molecules at a given test condition
The viscous flow regime (laminar flow) is typical with a Kn < 0.01, while slip flow (where molecular effects are noticeable) at 0.01 < Kn < 0.1. Transitional and molecular flows are observed with 0.1 < Kn < 10, and the gas transport can be described as only molecular flow when Kn > 10. In the viscous flow regime, the flow can be modeled based on the Hagen-Poiseuille laminar flow model (see eq 3). In the transitional and molecular flow regime, the flow can be modeled based on molecular phenomena (Knudsen model, see eqs 4–5). The micro-flow sensor calibration (or conductivity) is proportional to the mass flow (μg/min) where Kn > 0.55. The micro-flow sensor monitors its Kn range and uses its proper calibration coefficients.
The sensor calibration at vacuum level results in constant mass flow conductance, where its Kn > 0.55. Figure 2 shows a relationship between mass flow and electronic signal (called the sensor calibration curve) applying two different inlet pressures. The inlet pressure does not affect mass flow results as long as the sensor operates at these transitional and molecular flow conditions. Once calibrated, the sensor will measure mass flow and display flow rates at the configured flow rate units. The micro-flow sensor also includes an absolute pressure sensor (0–10 torr) to ensure that the test is performed at the pre-set Knudsen number and vacuum level, and a temperature sensor. Pressure and temperature measurements are also required to calculate gas density when converting mass flow to volume flow.
A typical micro-flow calibration curve of sensors as used for this study at Kn > 0.55.
To compare to other test methods it is convenient to convert the flow measurements units from μg/min to std cc/s, a commonly used unit for helium mass spectrometry measurement. Note that 0.1 μg/min of air at 1 atm absolute pressure and 20 °C (STP condition) is equal to 1.39 × 10−6 std cc/s (1200 μg/cc density of air at STP).
From Figure 2, the micro-flow sensor demonstrates the measurement sensitivity range of 0.1 to 10 μg/min or range of 1.39 × 10−6 to 1.39 × 10−4 std cc/s of air. Comparing this range to Kirsch's work (13), the mass flow measurement range is equivalent to the helium leak rates of 4 × 10−6 to 5 × 10−4 std cc/s, which corresponds to Kirsch's measurement of a 0.4–2 μm micropipette.
Testing Methods
Figure 3 shows a diagram of the mass extraction measurement system. The testing chamber, measurement sensor, and vacuum generation package are connected and controlled by valves. A sample is placed inside the testing chamber constructed of stainless steel. The testing chamber is designed to minimize the headspace air volume between the sample and testing chamber. Smaller test chamber volume results in faster time response. The CCI mass extraction instrument for sterile products will typically operate in the transitional and molecular flow regimes, where the container is exposed to a vacuum level of <5 mmHg absolute pressure. At that vacuum level, leakage from a liquid filled container will cause the liquid to boil (the boiling point of water at 20 °C is 17.54 mmHg) and hence enhances the material transfer.
Mass extraction instrument—test system diagram.
Materials such as paper labels contain a fair amount of moisture, which provides a significant amount of mass out-gassed under a vacuum. Therefore, paper labels should not be applied to samples for CCI testing. Furthermore, labels can cover defects. Other materials such as plastics and elastomers can also provide out-gassed mass under a vacuum. The out-gassed mass from the material itself can be minimized during evacuation and should not be significant in comparison to the mass from defects. In addition, mass entrapped between components should be evacuated prior to final mass flow measurement. Moisture in the environment can also affect mass flow testing due to moisture evaporating from the testing chamber surface. The method has been optimized to minimize measurement variations due to environmental relative humidity fluctuations during normal operation. If the evacuation time increases, background noise will be reduced proportionally. However, the evacuation should not be too long because mass can be depleted during evacuation if there is a defect in the air headspace of the sample. Therefore, appropriate evacuation time should be optimized. It was found that the 24 s evacuation time was sufficient to reduce the background noise. Also, it was experimentally determined that the mass extraction instrument using 24 s evacuation time was able to detect a 5 μm nominal laser hole in the air headspace of a sample (∼0.1 cc, ∼3 mm gap in 1 mL long syringes). If there is a larger defect, the defect can be detected during the large leak testing step. Or if the defect size is large enough to remove all air mass from the headspace during evacuation, the internal pressure can reach a vacuum level similar to the chamber pressure. This will result in liquid evaporation (boiling) and generation of a constant flow of liquid vapor which will be detected during the test.
The mass extraction test has four major steps to check the integrity of C/C systems. Testing time required for each step should be appropriately determined. Experimentally optimized testing steps have been used for many applications and proven by robustness studies. Examples of the test results from each step are shown in Figure 4.
An example of test profiles between intact and samples with a 2 μm defect.
-
Quick evacuation step: The chamber starts at barometric conditions. First, only the quick evacuation valve is open and the vacuum generation package draws a vacuum for the time required to reach the desired vacuum level and remove most of the air surrounding the C/C system. It was found that 2–3 s are enough for glass vials and syringes.
-
Large leak step: This is an initial quick check for any gross leak where the air is directed only through the sensor, while the vacuum generation source is isolated (the quick evacuation valve is closed). If there is a gross leak, the sensor detects the gross leak quickly by means of excess flow rate and/or pressure increase and the instrument will stop running immediately and close the test valves. It minimizes spills from a gross leak and avoids mass depletion and system contamination with liquid. This step easily detects defects of 10 μm and larger.
-
Evacuation step: All valves are open at this step and the vacuum generation package draws a vacuum from the test chamber through the quick evacuation loop and through the micro-flow sensor. If there is no gross leak, the instrument performs a long evacuation to minimize any background noise from moisture in the air and out-gassing from the container closure materials. The 24 s evacuation time has been determined for glass vials and syringes.
-
Measurement step (stability and test): All flow is directed only through the micro-flow sensor. The micro-flow sensor measures the mass flow. The mass flow continuously increases until it reaches a steady state. For CCI testing, it is not necessary to obtain the steady state mass flow. The goal of CCI testing is to distinguish the mass flow between intact and defective c/c systems. It was determined that 15 s is enough time to make the clear difference between intact and 2 μm-defect samples. Typical test results are shown in Figure 4.
Syringe samples can typically have 1–3 mm air headspace (∼0.1 cc). When a vacuum is applied to syringes, the plunger will move out when the break-loose force is smaller than the force created by the vacuum. If the plunger moves, the volume of the air headspace of the sample increases and the internal pressure decreases (ideal gas law), resulting in reduced total pressure difference between the inside and outside of the syringe. This results in less mass flow for a given defect size. Therefore, a plunger holder should be utilized to hold the plunger in place.
When the plunger holder holds the plunger in place, the plunger is stressed under a vacuum and the plunger may deform. After the plunger was restrained in the barrel, the stress applied to the plunger ribs was analyzed before and during the vacuum using a finite element analysis tool. The stress increased after a vacuum is applied but the increased stress is not significant. The analysis results also showed that risk of false passing results are minimal from the plunger deformation. This was analyzed with stress-strain models using a finite element analysis tool. It is recommended that a similar analysis should be considered for each configuration.
Materials and Equipment
Materials
Typical pharmaceutical grade tubing glass vials, rubber stoppers, and aluminum seals were used to prepare vial samples for this study. Tubing glass syringe barrels (1 mL long) with a staked needle, an elastomeric needle shield, and a plastic cover were used. FluroTec–coated and uncoated plungers were used to prepare the syringe samples for this study.
Drug product A is a solution formulation of a protein. The drug concentration is 3 mg/mL formulated in citrate buffer, mannitol, polysorbate, and other ingredients. Drug product B is another solution formulation of a protein. The drug B concentration is 120 mg/mL and formulated similarly to product A. A buffer solution was prepared with the same formulation without active product ingredient (API). A surrogate solution was prepared to mimic material properties of a 120 mg/mL drug product solution. Material properties between surrogate and drug product solutions are similar in terms of viscosity and surface tension, but the drug product solution contains protein.
Equipment
Two mass extraction instruments (Advanced Test Concepts (ATC), Inc., Indianapolis, IN) were used. Both instruments were built with micro-flow sensor measurement range of 0–10 μg/min and pressure measurement range of 0–10 torr (Part No. IM2-010U-10T-ME2).
X-ray μCT (computer tomography, Skyscan, Belgium) with a cone beam configuration equipped with a 10 megapixel X-ray camera was used to analyze geometries of defects. The system has a 5 μm focal spot size. Projection images were collected as the specimen rotated and were then reconstructed into a cubic array of volumetric pixels (voxels) with the same edge dimension as the projections. The reconstruction process was performed using NRecon software (Skyscan).
Defect Simulation & Fluid Theories
Defects need to be created consistently to demonstrate the method's capability and the created defects should be representative of real product defects. Real defects can be caused by components themselves such as cracks in glass or molding defects in plungers, improper sealing, or extrinsic materials such as fibers in a sealed area. There are several ways to simulate defects. Defects with different geometries will provide different types of flow rates. In this section, the geometries of several defects are analyzed first and fluid transport theories through micro-geometries are reviewed.
Glass Micropipette
The glass micropipette has been used predominantly to show the detection limit of CCI physical testing methods. The tip of a glass micropipette is very fragile, so it can be easily damaged during sample preparation. Therefore, its integrity should be verified microscopically before and after testing. Glass micropipettes are drawn and cut to get a known diameter, so the shape of the channel is tapered. The tolerance of the tapered tip inner diameter (ID) is typically ±20%. The length of the channel for a 2 μm glass micropipette is difficult to define but it is assumed to be shorter than 0.5 mm. Figure 5 shows an image of a glass micropipette tip. The actual channel is shown with the measurement and the glass wall is shown around the channel.
Image of the tip of a 2 μm glass micropipette (optical digital microscope ×4000).
Laser Drill
Another common way to create defects is to laser drill holes in c/c systems. For pharmaceutical sterile product CCI studies, laser drilling techniques are used to create an orifice greater than 5 μm nominal size directly in the glass or plastic container wall. A certain wavelength laser beam is applied to a point on the surface of the C/C system. It is absorbed and heat is generated on the point. The material can be melted and the hole is created. The drilling rate can be controlled but the depth of the hole cannot be controlled. Therefore, a desired hole size is created only by a trial and error approach.
The geometry of a laser-drilled hole is a tapered hole protruding partially into the side wall and results in micron cracks which create a tortuously irregular path into the container (14). It does not achieve the desired hole size on the surface of glass material. When glass material gains heat, cracks will be generated. If the laser beam is allowed to go through the entire thickness of the glass wall and if a desired diameter size is achieved, the flow rate from the entire hole will be significantly higher than expected because cracks will exist around the hole. In order to get the size of holes equivalent to 5–10 μm glass micropipettes, the drilling should be stopped in the middle of the glass wall and cracks from the drilling will provide a flow rate equivalent to the target defect size. Figure 6 shows X-ray microcomputer tomography (X-ray μCT), digital microscope, and SEM (scanning electron microscopy) images for a 5 μm nominal defect created by a laser drilling technique. X-ray μCT data consisted of 1048×2000 pixel projections which were collected as the specimen was rotated 180 degrees in 0.1 degree increments. The pixel size, projected from the specimen, was 1.9 μm. Note the resolution of the system is ultimately determined by the focal spot size and not the pixel size.
Images of a 5 μm nominal laser-drilled hole in glass wall [(a) X-ray image, cross-section in XY plane, (b) cross-section YZ plane, (c) optical digital microscope image (×200), (d) SEM image (×2400) of a portion of image (c)].
Images from cross sections in x and z directions were visually inspected. There was no through hole equivalent to 5 μm found. Figure 6 (a) shows the cross section in the XY-plane approximately in the center of the hole. Figure 6 (b) shows an image in the YZ plane indicated by the vertical line in 6 (a) showing cracks of the glass wall. Figure 6 (c) shows an optical microscope image of the inner surface of containers and (d) shows its SEM image. The star-pattern fracture does pass through the wall and the crack is roughly 1 μm at the widest point.
As described above, it is difficult to define the geometry of laser-drilled holes given that they are actually a series of cracks. The vendor developed a calibration curve between nominal defect size and flow rate. The nominal size of defects was determined based on their calibration curve and a certificate of calibration flow was provided. It was found that laser drilled defects smaller than 10 μm show somewhat large variations, as will be discussed in Results and Discussion section. This is expected based on the analysis above.
Wire
When a wire is placed in a rubber compressed seal area such as a stopper or plunger seal, a channel can be created around the wire. The size of the formed channel will depend on the diameter of the wire, the formulation of the rubber, the design of the plunger/stopper, compression force, etc. For example, if a wire is placed in the interference fit between a glass barrel and plunger in a syringe, the rubber plunger is compressed against the rigid glass barrel and wire. This results in a gap between the glass barrel, wire, and compressed plunger. The rubber cannot be deformed completely to fill the gap; thus, a certain size channel can be created. X-ray μCT was used to analyze channels in the interference fit. Figure 7 (a) shows the cross section in the z direction of the plunger and wire—a 150 μm outer diameter (OD) microtube—inserted into the interference fit. As noted in the image, the plunger has three ribs. When the plunger is compressed, a different amount of stress is exerted on each rib. It was found that the bottom rib can be compressed the most due to the geometry of the plunger. As a result, the channel in this rib is the smallest. Figure 7 (b) shows the smallest channel from cross sections in the XZ plane of the inserted plunger. Figure 7 (c) shows channels identified by the software, which were analyzed to estimate the total surface area of channels. The total surface area represented by these channels was determined to be approximately equivalent to that of a 10 μm diameter circle.
X-ray CT images—150 μm wire.
Next, a smaller OD wire (80 μm OD metal-coated human hair) is inserted in the interference fit. The metal coating generates a sparkling image in X-ray, helping to distinguish the wire from the plunger as can be seen in Figure 8. X-ray μCT could not identify any channels. X-ray μCT data consisted of 2096 × 4000 pixel projections which were collected as the specimen was rotated 180 degrees in 0.15 degree increments. The pixel size, projected from the specimen, was 4.2 μm. Even with a focal spot specification of 5 μm, the image resolution can be limited by poor contrast and the signal to noise ratios of the images. The minimum size gap that could be discerned was 10 μm, so it can be concluded that there are no channels greater than 10 μm resulting from introduction of an 80 μm hair.
X-ray CT image for approximately 80 μm hair (metal coated) inserted in the interference fit.
Microtube
The microtube is manufactured from silica and it is externally coated with a polymer. It is strong but flexible, so it can be easily handled to minimize any breakage during sample preparation. The geometry of the microtube is a circular cylinder with a ± 1 μm ID tolerance. A 10 cm length 20 μm (ID) microtube was scanned with an X-ray μCT. The ID was determined to be consistent and no blockage was found.
The microtube can be installed conveniently through the body of the plunger and stopper with little concern about breakage during sample preparation. The shortest length microtube that can be practically installed in syringes and vials is about 5 mm. This defect length does not represent any potential real-life defects. Based on the mathematical models (eqs 3–5), it is clear that increased length will reduce the flow rate. In addition, it was experimentally found that 5 mm length microtubes with 2, 5, and 10 μm diameters are more likely to become clogged with drug product, due to the lengthy channel, compared to a glass micropipette and laser-drilled defects. (Note: see “Results and Discussion” for the effect of product “clogging.”)
Fluid Transport through Micro-Geometries Theories
Analytical fluid models require defined geometry dimensions with minimal entry and exit flow pressure losses. However, as described, the micron size of simulated defects cannot meet these requirements. Therefore, there is no perfect model for each type of defect. Fluid models should be used for “magnitude of order” estimation and analysis of the parameters influencing the material transport phenomena. Various fluid models have been previously reviewed (10, 15, 16).
The fluid models for micron-size channels can be divided into models for compressible fluid (gases) and incompressible fluids (liquids). To describe the behavior of fluid resistance to flow (or pressure loss) through a micro-geometry (defect) with liquid on one side and with air on the other, the Young-Laplace Equation (eq 2) can be used. The equation can estimate how much total pressure difference is required to remove the air lock and to break the liquid surface tension to initiate fluid transport.
where ΔP is the required pressure to overcome the air lock, γ is the surface tension of the liquid, θ is the contact angle, and D is the flow path diameter (or hydraulic diameter).
The surface tension (γ) of Product A liquid solution is ∼50 mN/m. Note that the surface tension of water is ∼72.8 mN/m at 20 °C and the contact angle is near zero (15). Total pressure differentials of 94 kPa, 38 kPa, and 19 kPa are required to remove the air lock for 2, 5, and 10 μm diameter defects (e.g., glass micropipettes), respectively. If this pressure difference does not exist across the defect no flow can occur in a case of liquid leakage into air. The pressure to break the surface tension is independent of the flow path length. However, if a lengthy microtube is used, it is difficult to achieve the required pressure difference across the outlet of the microtube due to pressure loss from inlet to outlet. Therefore, microtubes should not be used to demonstrate detection in the 2–5 μm range.
Another type of fluid model, the Hagen-Poiseuille model (eq 3) can be used to describe non-compressible (liquid) laminar flow in the viscous flow regime that is derived from total pressure difference, capillary path geometry, and gas properties. This model is limited by low Reynolds number flows (laminar flow) and larger L/d (length to diameter) ratio. This equation can be also utilized to calculate the total pressure difference required to force liquid flow.
where Q is the volumetric flow rate, d is the diameter of the channel, ΔP is the pressure differential, L is the length of the channel, and μ is the dynamic viscosity of the fluid. Eq 3 is applicable mostly for non-compressible fluids (liquid).
For a compressible fluid (gases), Knudsen number, Kn, can be used as an indicator of the flow regime. Kn < 0.01 represents continuum (viscous) flow with negligible slip effects and Kn > 10 represents the molecular flow regime (16). For inert gas flow, a generic model, based on gas flow through a micro channel at any Kn and free molecular flow for low pressure differential, was suggested by Knudsen (Equations 4–5).
where ṁKn is the mass flow at any arbitrary Knudsen number, ṁfm is the mass flow rate in the free molecular flow regime, Kn is the Knudsen number averaged along the flow path, R is the specific gas constant (universal gas constant divided by the molecular weight of the gas), and T is the absolute temperature. This model provides better results than the Hagen-Poiseuille model when applied to inert gas or vapor flow through micro-channels (e.g., 2–5 μm glass micropipette and 5–10 μm laser drilled cracks).
Experimental Designs
Defect Sample Preparation
In order to determine the detection limit of the method and understand the method's capabilities, various types of defects were created. Glass micropipettes (2 and 5 μm ID (±20%), World Precision Instruments, Inc., Sarasota, FL) were installed through vial stoppers using a 16G needle. An epoxy sealant was applied around the punctured area to secure the seal. It is difficult to install a glass micropipette through a syringe plunger and then insert the plunger into the glass barrel without breaking the micropippettes. Therefore, the data generated for rubber-stoppered vials was used also to represent defective syringes. When a glass micropipette is submerged into a solution, the solution may penetrate through the micropipette even if some effort was put into minimizing the capillary effect. Before and after testing, the samples' tips were inspected under a microscope to ensure the tips were intact. Also, when glass micropipettes were installed for powder-filled vials, in order to avoid breaking the micropipettes' tips these micropipettes were not submerged directly into the powders. Vials were positioned horizontally enabling the powder to spread out on the side wall of the vial. The micropipettes were inserted into the middle of the vials, then the vials were placed in the vertical position, submerging the micropipettes' tips in the powder.
Laser-drilled cracks for vials and syringes were prepared by Lenox Laser, Inc. (Glen Arm, MD). The vendor measures the leak rate with laser-drilled samples to verify the nominal size of the holes. The smallest laser drilled crack has a leak rate equivalent to that of 5 μm nominal size. A variety of sizes of nickel-titanium alloy wires (25, 38, 50, and 75 μm OD, Dynalloy, Inc., Tustin, CA) were also installed between the stopper and crimp seal in the vials and in the interference fit between plunger and inner glass syringe wall. The diameter size was verified using a microscope. These defects (glass micropipette, laser drilled crack, wire) were prepared for air, powder, water, placebo solution, surrogate solution, product A (3 mg/mL), and product B (120 mg/mL) solution-filled vials or syringes.
Design of Experiments
Experiments were statistically designed to understand the detection limit and robustness of the method for both vials and syringes. The detection limit study was intended to verify what size defects can be characterized by the method with a defined probability of detection. The robustness study was designed to demonstrate the reliability of the mass extraction measurement with respect to deliberate variations in method parameters.
For the detection limit study, various types of defect samples were prepared as explained in defect sample preparation and included different types of contents. If there is a defect in the C/C system and if the defect is in contact with the contents of the container, this will affect the mass flow testing results due to factors such as surface tension, viscosity, molecular weight/size, etc. Therefore, these factors were included in the studies to determine the detection limit of the method.
For the robustness study, multiple operators, testing days, and two stand-alone instruments were used to understand variations of the method. A total of 12 different combinations including two extreme vial sizes (2 and 50 mL) were incorporated into the design and 6 vials per combination were tested. For the syringe study, a total of 8 combinations were incorporated into the design and 90 syringes per combination were tested. From previous characterization studies, it was found that mass flow results from the syringe container closure system showed a larger variation than did the vials, so more samples were included for the syringe robustness study. A known size defect (2 μm glass micropipette) was installed in a fixture connected to the testing chamber. This setup was intended to simulate a 2 μm defect located in the air headspace of the samples. Because the limited headspace can reduce flow rate through certain defect sizes, it was verified that the mass flow results are similar between the 2 μm glass micropipette installed in the 2 mL vial's headspace and the 2 μm glass micropipette installed in the test circuit (Figure 5). There is a switch that can control the opening and closing of the valve between the fixture and testing chamber. This enabled the use of one micropipette size for testing of all samples for the robustness study. This setup can eliminate glass micropipette breakage and variation from different micropipettes. This “built-in” 2 μm micropipette design is also beneficial for routine system suitability checks and periodic system performance verification.
Results and Discussion
Prior to instrument runs to test the samples, instrument suitability was assessed to ensure the instruments were operating correctly and able to detect a leak from a 2 μm glass micropipette. Ten measurements from the empty chambers and five measurements from the 2 μm glass micropipette were required to meet the instrument suitability criteria. The instrument is designed to measure the mass flow between 0 and 10 μg/min. Any leak greater than 10 μg/min is considered as a gross leak. The gross leak is indicated with the maximum value of 10 μg/min.
Correlation between Defect Size and Mass Flow
The correlation between defect size and mass flow was determined using the glass micropipette and the testing setup shown in Figure 4. Various size glass micropipettes were installed in vials (empty) and mass flow was measured—five each of intact, 0.2, 0.4, and 1 μm samples, and thirty 2 μm, three 5 μm, and three 10 μm samples. Figure 9 shows mass flow results as a function of the size of glass micropipette. There is a clear correlation between mass flow and glass micropipette size. Mass flow results from 0.2, 0.4, and 1 μm glass micropipettes are not much different than intact samples but the mass flow results from 2 μm glass micropipettes show a clear difference between intact samples and samples with 2 μm glass micropipettes. As mentioned, there is a ±20% size variation on each size glass micropipette, so the mass flow shows some variation in results from thirty individual 2 μm glass micropipettes. This correlation can be used to approximate the size of defect channels between 0 and 5 μm, or identify a defect size larger than 5 μm.
Mass flow from various size glass micropipettes installed in empty vials.
Detection Limit
Based on the data discussed above, it was determined that the mass extraction system is capable of detecting a leak from a defect equal to or greater than 2 μm diameter using a glass micropipette. Next, in addition to the glass micropipette defects, other types of defects were created in an effort to determine the detection limit. First, various sizes of laser-drilled cracks in syringes and vials were tested, including 20 intact, eighty 5 μm, forty 10 μm, and eighty 15 μm samples. Figure 10 shows that the mass extraction instrument is capable of detecting laser drilled cracks greater than 5 μm nominal size. Laser-drilled cracks are irregular, so the mass flow results from 5 μm nominal size laser-drilled crack samples show relatively large variation.
Mass flow from various nominal size laser-drilled cracks in empty samples.
Another defect type created in this study was a wire placed in compressed rubber seals. Figures 11–13 show mass flow results from various size wires placed in the interference fit of syringes and along the vials' stopper-plugs and neck-stopper compressed seals. As explained, X-ray μCT did not identify any channels around the compressed uncoated plunger and 75 μm (OD) wire. Figure 11 shows that mass flow increases as wire size increases. Mass flow results from the 50 μm wire show a broad range of mass flow results between 2.3–10 μg/min. It means these samples may have channels ranging approximately between 2 and 10 μm. All samples with 75 μm wires show high mass flow but no channels were identified using X-ray μCT, so it is assumed that these samples have channels smaller than 10 μm. Figures 12 and 13 show mass flow results for FluroTec-coated plungers in syringes and aluminum crimp-sealed stoppers in vials. Different mass flow results are expected due to the different design of plungers and different types of c/c systems. However, the results showed a similar trend as compared to the uncoated plunger interference fit. A transition phase can be seen in the example of a 38 μm wire placed in a coated plunger in interference fit and stopper compressed seals. In this phase, it is assumed that the size of the created channels ranges between 2 and 10 μm.
Mass flow from various size wires in interference fit of syringe barrel and uncoated plunger B (empty syringes).
Mass flow from various size wires in interference fit of syringe barrel and FluroTec coated plunger A (empty syringes).
Mass flow from various size wires in glass vial/rubber stopper/aluminum crimp seal (empty vials).
This data demonstrates that the mass extraction system is capable of detecting various types of defects when the defect is exposed to air. Next, a series of studies was conducted to determine the impact of measuring mass flow in the presence of different contents such as powder or liquids (e.g., water, placebo, and product). If powders do not block defects and if liquids can flow through defects under given conditions, the mass extraction instrument will measure the flow rate. Table I shows mass flow results from studies conducted with various types of defects and contents. All defects were exposed directly to the contents.
Mass Flow Results for Defect Samples (# of Positive Results/# of Samples)
Figure 14 shows all individual mass flow results for the samples shown in Table I. Defect samples containing mannitol powders showed high mass flow results. It means powders with sizes and properties similar to mannitol do not interfere with mass flow nor impede the ability to detect defects. Note that some of the glass micropipettes are assumed to have been broken since mass flow results from some of the 2 μm glass micropipettes are higher than the normal variation range observed in earlier studies. The integrity of glass micropipettes inserted into powder filled samples cannot be easily verified microscopically. It was also found that water, placebo, and low concentration drug product can flow through 2–5 μm-size defects. For these samples, micropipette tips were inspected to verify the intact tip and that the solution migrated through the micropipette both prior to and following testing for mass flow. This data also suggested that in the presence of high concentrations of certain drug products, it can be difficult to detect flow through 2–5 μm-size defects. Highly concentrated protein drug products are typically viscous solutions. In an effort to confirm the theory that high viscosity interfered with mass flow, a surrogate solution that has a similar viscosity was also tested. The experiment showed that the surrogate solution defects were detected using 5 μm glass micropipettes. From this experiment, it can be concluded that high viscosity alone was not a factor in impeding the flow through 5 μm-defect channels. The other potentially contributing factor was the active ingredient, which in this case was a large molecule protein. It has been hypothesized that the large molecules can agglomerate around or in the defect channel and thus hinder the flow of solution. This clogging effect potentially occurs with high drug concentrations of large molecules and possibly during storage of any proteins. Further studies need to be conducted to better understand this relationship and its impact on CCI testing methods for high concentration/large molecule drug products.
Mass flow results for defect samples.
Robustness
Results from the robustness studies were analyzed using JMP|Pr release 7. For vials, it was found that the variation of mass flow is low (0.05 μg/min standard deviation) and there is a clear difference between intact vials and vials with a 2 μm defect (glass micropipette). An acceptance limit of 2 μg/min was determined using six standard deviations of the mass flow results from the 2 μm defect vials. Figure 15 shows individual mass flow results with the acceptance limit.
Mass flow results between intact and samples with a 2 μm defect for syringes and vials (360 data points/column for syringes and 36 data points/column for vials).
For syringes, it was expected that the overall variation would be larger than vials, so more samples were included in the robustness study. Syringes are a more complex design and have more plastic and rubber materials that can provide outgassing. There is approximately 0.1 cc of air in the needle shield cavity and the needle shield is designed to release the air under a certain pressure. It is expected that the entrapped air is removed during evacuation under a vacuum. If there is remaining air, it might create a mass flow during the test. In addition, it is assumed that the syringe in the testing chamber may move during testing. If this happens, the mass flow can suddenly change but not significantly. An analysis of the mass flow results shows a large, statistically significant difference between intact syringes and syringes with 2 μm defects. The analysis also shows that the differences between operators, instruments, and days are each statistically significant at the 5% level, but lot-to-lot differences are not. The operator, instrument, and day differences are small relative to the mass flow differences between intact syringes and syringes with 2 μm defects. Therefore these other factors are deemed not to be practically significant. Figure 16 shows individual mass flow results for all combinations. All results were pooled together for analysis. An acceptance limit was determined using three standard deviations of the mass flow results from 2 μm-defect syringes, which is 2.8 μg/min at the given test time.
Mass flow results from the robustness study of syringes.
The robustness studies demonstrated the reliability of the mass flow measurement with variations in method parameters. The determined mass flow acceptance criteria can be used to ensure the CCI for syringes and vials, based on a limit of detection equivalent to a single breach of greater than 2 μm in the C/C system.
Conclusion
The mass extraction measurement instrument operating in the transitional and molecular flow regimes demonstrated that a 2 μm defect (glass micropipette) can be reliably detected. A variety of defect samples were created and various types of contents were filled into the defect samples. The mass extraction instrument successfully and reliably detects 2 μm-size defect samples containing air, water, placebo solutions, and mannitol powder. However, when the defect is exposed to a drug product containing large molecules, the solution does not quickly flow depending on concentration. It was found that low concentration drug solution (3 mg/mL) flowed quickly but high concentration drug solution (120 mg/mL) could not flow through 5–10 μm glass micropipettes under a vacuum. It is hypothesized that large molecules agglomerate around or in the defect channel. Further studies need to be conducted to understand the fluid properties of various drug solutions and develop appropriate CCI testing methods that may potentially overcome this clogging phenomenon.
The mass extraction instrument also demonstrated robust results for both syringes and vials. Using 3 or 6 standard deviations, an acceptance limit was determined with 2 μm-defect samples (glass micropipettes). Therefore, this mass extraction measurement technology is a valuable method to confirm the integrity of C/C systems for qualification, sealing processes qualification, and any other applications in which CCI needs to be checked.
Conflict of Interest Declaration
The authors declare that they have no competing interests.
Acknowledgments
The authors acknowledge Randy Thackrey, David Knoy, Sarah Demmon, Mark Claerbout, Mark Strege, Mary Stickelmeyer, Ross Allen, Yu Hu, Tim Kramer, Eric Adamec, Craig Kemp, Michael Foubert (from Lilly), Rob Chirico, and Alex Ivanchenko (from ATC) for their technical support and guidance.
- © PDA, Inc. 2012
Jump to section
Related Articles
Cited By...
- Container Closure Integrity Test Using Frequency Modulation Spectroscopy Headspace Analysis with Carbon Dioxide as a Tracer Gas
- Comparing Physical Container Closure Integrity Test Methods and Artificial Leak Methodologies
- Container Closure Integrity Testing--Method Development for Freeze-Dried Products Using Laser-Based Headspace Oxygen Analysis
- Method Development for Container Closure Integrity Evaluation via Headspace Gas Ingress by Using Frequency Modulation Spectroscopy
- Container Closure Integrity Testing--Practical Aspects and Approaches in the Pharmaceutical Industry
- Artificial Leaks in Container Closure Integrity Testing: Nonlinear Finite Element Simulation of Aperture Size Originated by a Copper Wire Sandwiched between the Stopper and the Glass Vial
- Integration of Regulatory Guidelines into Protein Drug Product Development
- Feasibility of Using Fluorescence Spectrophotometry to Develop a Sensitive Dye Immersion Method for Container Closure Integrity Testing of Prefilled Syringes