Commercialization of protein-based therapeutics is a challenging task in part due to the difficulties in maintaining protein solutions safe and efficacious throughout the drug product development process, storage, transportation and patient administration. Bulk drug substance goes through a series of formulation, fill and finish operations to provide the final dosage form in the desired formulation and container or delivery device. Different process parameters during each of these operations can affect the purity, activity and efficacy of the final product. Common protein degradation pathways and the various physical and chemical factors that can induce such reactions have been extensively studied for years. This review presents an overview of the various formulation-fill-finish operations with a focus on processing steps and conditions that can impact product quality. Various manufacturing operations including bulk freeze-thaw, formulation, filtration, filling, lyophilization, inspection, labeling, packaging, storage, transport and delivery have been reviewed. The article highlights our present day understanding of protein instability issues during biopharmaceutical manufacturing and provides guidance on process considerations that can help alleviate these concerns.
The term “formulation, fill and finish” refers to the series of processing steps that are needed to turn a purified drug substance into the final dosage form, the finished product, for the market (1). The formulation step involves taking the purified protein at the desired concentration and dispensing it with the correct excipients that can ensure product quality and integrity during the subsequent fill/finish steps including filtration, filling, lyophilization, packaging, storage, transport and delivery. A robust formulation would need to keep the biopharmaceuticals stable not only during shelf storage but also during these manufacturing steps. At the same time, key operating and process parameters should be optimized to obtain a robust manufacturing process. The problems for protein therapeutics could be very different from the traditional small-molecule pharmaceutical processing and may require special handling and storage conditions to ensure product quality (2, 3). For instance, protein thermal instability is one of the main reasons why protein drugs need to be maintained under cold temperatures during storage and transport to achieve longer shelf life. Similarly, other stresses such as photo exposure and mechanical agitation could also impact the stability of protein products.
Proteins are large macromolecules made up of a sequence of amino acids and characterized by a unique three-dimensional structure corresponding to their biologically active state. The native structure of a protein molecule is the result of a fine balance among various interactions including covalent linkages, hydrophobic interactions, electrostatic interactions, hydrogen bonding and van der Waals forces. Intraprotein and protein−solvent interactions both play an important role in maintaining the protein structure and its stability. The free energy of unfolding has been generally reported to be quite small, in the range of 21−63 kJ/mol (2). Since the folded state of protein is only marginally more stable than the unfolded state, any change in the protein environment may trigger protein degradation or inactivation.
The degradation pathways for protein therapeutics are many and often complex. Simplistically speaking, these pathways can be divided into two categories: physical degradations, which do not involve covalent bond modifications, and covalent modifications (for reviews, see refs 1−5). Physical degradations are most commonly manifested by protein aggregation. This type of degradation involves assembly of monomeric units of proteins, and dimerization is a common occurrence within these set of events (6). Higher order protein oligomers are often referred to as “high molecular weight species” or protein aggregates (4−8). These protein aggregates can be either soluble or insoluble. Recent evidence has suggested that protein aggregation occurs by a specific association of partially denatured polypeptide chains, as opposed to nonspecific co-aggregation (7, 8). Protein aggregation can be assessed by a variety of techniques, such as size exclusion chromatography, field flow fractionation and analytical ultracentrifugation for soluble aggregates (9) and light obscuration/scattering techniques for insoluble aggregates (10). A second type of protein degradation is a change in the secondary, tertiary or quaternary structure of the protein, which does not involve protein−protein interactions. Biophysical techniques, such as circular dichroism, FT-IR and fluorescence are usually employed to assess such structural changes. These two degradation pathways can be intimately linked: a change in protein structure often precedes protein aggregation phenomena (4). Native aggregates are those in which there is only an assembly of protein monomers without a change in structure, whereas there is a change in protein structure (secondary and/or tertiary) in the formation of non-native protein aggregates.
There are several possible covalent modifications in proteins. One common modification is protein fragmentation, which involves the cleavage of a peptide bond. Residue specific modifications include but are not limited to aspartate isomerization, protein oxidation, deamidation, pyroglutamic acid formation and disulfide bond shuffling (5). Many times, proteins undergo post-translational modifications, such as glycosylation. Changes in glycosylation patterns under storage and processing conditions are also known to occur (11). Further, formulation excipients sometimes have the potential to interact with protein side chains, such as the glycation reaction between reducing sugars and side-chain or N-terminal amino groups (12). Covalent degradations that lead to changes in net charge of the protein can be captured by ion exchange chromatography (13, 14) and capillary isoelectric focusing (cIEF) (15). Weak cation exchange chromatography is commonly used to monitor antibody stability. Techniques combining reverse phase and mass spectrometry, such as peptide mapping, are more comprehensive and have the potential to detect most, if not all, covalent modifications (13, 16), but these are more time-consuming and resource intensive.
This review article presents an overview of various formulation, fill and finish operations. The key aspects of processing steps that can affect stability and integrity of a product are discussed. Figure 1 presents the series of operations and various factors that can impact product quality. In the sections that follow, each unit operation is described in detail. The impact of some of the key operating input parameters on protein stability is included, and guidance is provided on scale-down studies needed to evaluate such destabilizing factors. For many fill and finish operations, such as freeze−thaw, mixing, and filtration, the main concern would be physical stability of the protein. However, exposure to light and various manufacturing equipment surfaces could trigger covalent modifications as well. Therefore, orthogonal assays to determine the physical and covalent stability of the molecule should be carried out to determine the overall stability of the molecule to various bioprocessing stresses. In addition to formulation and other unit operations, protein instability issues may arise from interactions of drug product with packaging and components of delivery device. Current understanding of these challenges is included. Photodegradation of light-sensitive products is also discussed in a later section.
Formulation and Fill-Finish Operations
Bulk Freeze−Thaw: Advantages. Bulk freeze−thaw is commonly employed during biopharmaceutical manufacturing to gain operational flexibility while maintaining product quality. A frozen drug substance provides several advantages over liquid storage, including increased product stability, reduced possibility of microbial growth and alleviation of foaming issues during transportation, thereby eliminating the need to perform transport validation. Lowering the temperature to achieve a frozen bulk reduces the rates of degradation reactions and also immobilizes the protein molecule in a frozen matrix, thereby minimizing diffusive collisions that lead to aggregation. Lack of availability of free water also prevents several degradation reactions that are assisted by water, such as peptide bond hydrolysis and aspartic acid isomerization phenomena, further increasing the stability of frozen bulk in comparison to aqueous formulations. This greater assurance in product quality provides flexibility to schedule formulation-fill-finish operations based on the needs of the manufacturing facility. In this scenario, bulk drug substance is stored frozen and when needed is transported to the fill site where it is thawed, initiating a series of formulation and fill/finish steps. The application of freeze−thaw is not limited to storage of drug substance but is also used for the storage of pharmaceutical intermediates and formulated drug products.
Protein Freezing: Stability Challenges. While bulk freeze−thaw offers numerous operational and product quality benefits, it may also prove detrimental to protein stability. Cryoconcentration is one of the common mechanisms through which protein destabilization could occur during freezing (17−19). As the freeze-front moves during the freezing process (Figure 2), the excipients as well as the proteins get excluded from the ice−liquid interface. As a result the concentration of the liquid bulk (yet to be frozen) close to the ice crystals increases progressively with freezing. Such concentration build up of excipients may result in changes in protein structure. Freezing of buffer solution can also cause change in pH due to selective precipitation of buffer components, which can also result in protein destabilization (20). At the same time, increase in protein concentration also increases the possibility of molecular collisions and may result in protein aggregation or precipitation. The extent of cryoconcentration is maximized if the rate of freezing is slow. As a result, uncontrolled freeze−thaw processes, where the freeze front velocities are lower, are impacted to a greater extent by the destabilizing effects of cryoconcentration. One way of minimizing such freeze concentration effects is to reduce the freezing times by increasing the heat transfer from the container. Dendritic ice growth is also preferred in order to minimize cosolute exclusion during freezing. This can be achieved by establishing directional heat flow and avoiding mixing during freezing. Mixing could be detrimental as it would suppress dendritic ice growth, making the ice−liquid interface more flat, and therefore result in increased cryoconcentration.
Proteins could also be susceptible to spontaneous unfolding at cold temperatures, referred to as “cold denaturation” (21). This effect is primarily attributed to the weakening of the hydrophobic effect with decrease in temperature. The thermodynamic justification of the cold denaturation temperature can be explained by the parabolic shape of the Gibbs free energy function as shown in Figure 3. A favorable negative free energy of unfolding favors thermal denaturation at higher temperature (Td). At lower temperatures, it is possible that ΔGunfolding may become negative again below a certain critical temperature (Tc), resulting in protein unfolding. Such cold-induced denaturation phenomena, though rare, have been reported for certain proteins (22, 23). Further review of this phenomenon can be found in the literature (21).
Very fast freezing rates can also prove to be detrimental to proteins (24). During freezing, protein molecules can concentrate and get unfolded (25, 26) on the ice−water interface, implying loss in protein activity. When freezing rates are very fast (e.g., submerging container in liquid nitrogen (17)), smaller ice crystals are formed and result in a large ice−liquid interfacial area (25, 26). Increased protein aggregation and decreased activity have been reported for liquid-nitrogen-based freezing systems (27, 28). Fast freezing can also trap air that would be released during thawing and may cause protein denaturation on air−liquid interfaces (29, 30).
Thawing Frozen Protein Solutions: Stability Challenges. Frozen bulk needs to be thawed before it can be formulated and processed. Thawing can cause further stress and damage to the protein. Slow thawing rates can result in ice recrystallization with small ice crystals growing into larger ones. Proteins may get denatured at ice−liquid interfaces and lose their activity (25, 26). Cryoconcentration created during freezing can further harm the protein during thawing. Like freezing, faster thawing rates are usually preferred for protein stability. While mixing during freezing may be detrimental to proteins, appropriate mixing during thaw is the key to minimize recrystallization and cryoconcentration related effects. Very slow mixing would contribute to longer thaw times and also would not be able to homogenize the solution (concentration gradients from the freeze step would continue to exist and may further increase). On the other hand, very high mixing rates would result in protein shearing, excessive foaming and plausible protein denaturation on the air−liquid interface. The mixing parameters should therefore be optimized to enhance thawing without affecting product quality.
Freeze−Thaw Technologies And Process Scale-Down. The challenges faced during the freeze−thaw process would be dependent on the technology employed for the large-scale process. Most of the stability issues discussed above occur when very slow freeze−thaw rates are applied, which is usually the case for uncontrolled rate technologies (17). For example, polycarbonate carboys (10 L and 20 L) are commonly used to freeze and transport bulk drug substance. Freezing is conducted by placing these carboys in walk-in or upright freezers at −30 or −80 °C. Since the path lengths are large and the heat flux is slow, the process times for the freeze−thaw operations could be very long as shown in Table 1. As a result cryoconcentration becomes an important factor governing product quality in these containers. Controlled rate technologies such as Celsius Paks or Cryovessels, on the other hand, can achieve faster freezing and thawing rates by using a combination of small path length and increased flux for heat transfer (18). Table 1 shows a comparison of these process times for the uncontrolled rate (carboys) and controlled rate (Celsius Pak) technologies. It has also been shown in literature that the extent of cryoconcentration is minimal for Celsius Paks (18).
Table Table 1.. Comparison of Freeze−Thaw Parameters for Carboy and Celsius Pak
carboy (10 L)
Celsius-Pak (16.6 L)
a Freeze time refers to time taken by the solution to go from +3 °C to −5 °C.b Freeze front velocity c Thaw time reported in the table refers to the time needed to thaw 8.5 kg of protein solution in a 10 L carboy at 2−8 °C.
freezing timea (h)
17.1 ± 0.9
1.6 ± 0.2
7.3 ± 0.4
25.5 ± 2.5
thaw time (h)
150 ± 15c
2.5 ± 0.5
solution homogeneity after thaw
The effects of bulk freeze−thaw on the product are protein-specific. It may not affect product quality for some protein solutions but may have negative effects on others. As a result, prior to large-scale processing, each product should be evaluated for the impact of multiple freeze−thaw operations on product quality. For early-stage products where product availability may be limited, scaled-down studies can be performed to mimic large-scale freeze−thaw process. For uncontrolled freeze−thaw processes, usually a smaller bottle with (a) surface area to volume ratio similar to that of the large scale and (b) material of construction identical to the large-scale container can be used. Freeze−thaw profiles from large-scale processes can also be mimicked on small scale using a controlled rate freezer. However, certain phenomena, such as cryoconcentration, could be process-scale-dependent and difficult to mimic in a small container. It is usually feasible to mimic the impact of protein−container interactions during freeze−thaw in smaller-scale experiments.
For controlled rate technologies, the freeze−thaw process is often scalable in terms of freeze−thaw times and heat transfer path length. For example, the path length for 30 mL, 100 mL, 8.3 L and 16.6 L Celsius bags is identical (42 mm) and helps make the process scalable. The heat transfer fluid temperature profile over time can be programmed to achieve the same freeze−thaw profiles at all scales. Figure 4 shows how the thermal control unit of the Celsius technology can be used to obtain similar freeze−thaw profiles for lab-, pilot- and full-scale systems. This provides the flexibility to conduct stability characterization studies at lab scale with very limited material. While such disposable bag technology offer numerous advantages with the freeze−thaw process, the final impact on the product quality is also governed by the impact of the container. Issues such as increase in protein concentration associated with the loss of water vapor from plastic bags have been reported for prolonged storage at room temperature (31). The product-packaging interaction, robustness of the bag's mechanical properties, permeability of the bags and the level of leachables and extractables should be characterized in detail to ensure that no impact on product quality over a period of storage time is observed.
Formulation Step. The first step after thawing the bulk is to formulate it with the right buffer and to the target concentration. The formulation step involves adding the desired excipients at target concentration and adjusting pH, conductivity and protein concentration (32). The final dosage form for the drug product could be different from that of the bulk drug substance. Sometimes it is operationally more favorable to store a drug substance at a higher concentration than the drug product, and therefore a dilution step would be needed during formulation. In other cases, a buffer exchange may be required between drug substance and drug product. To perform buffer exchange, a UF/DF step may then be required. There can be logistical challenges with the implementation of this step, such as whether to perform it at the bulk manufacturing site or at the drug product fill and finish site. One of the main challenges during the UF/DF process is arriving at the target bulk pH at the end of operation. Recent work by Stoner et al. have provided the groundwork for this phenomena and have provided a mathematical tool to calculate how much pH adjustment to make prior to the step to hit the target pH at the end (33). Similarly, a concentration step may also be needed if the drug product is formulated at a concentration higher than that of drug substance. High product concentration and viscosity could pose further challenges to membrane filtration during the concentration step. Impact of filtration on product is further discussed in the next section.
The purity of the excipients could be another key factor affecting product quality at this step (34). Certain impurities in the raw materials can trigger degradation reactions. Using animal-derived excipients may carry a risk of causing TSE (Transmissible Spongiform Encephalopathies), and this would need to be carefully evaluated. Exposure to different surfaces including tubing and tanks could also affect protein stability. Surfactants, such as polysorbate 20, added to the formulation buffer as stabilizers, could get adsorbed on these surfaces and lower the surfactant concentration, resulting in protein destabilization. Leachables and extractables (especially if disposable bags are being used) from the various contact surfaces also have the potential to affect product integrity.
Insufficient mixing during the addition of excipients could alter product quality due to solution inhomogeneity and also result in the final drug product not being able to meet its specifications. Excessive mixing, on the other hand, could create large shear stress that can denature proteins. Physical instability of proteins arising from mechanical stresses such as stirring and shaking in presence of various contact surfaces has been widely reported in literature (35−39). Air−liquid interfaces created during the mixing and pumping processes are another source of protein denaturation. Pumping can also result in addition of foreign particles in the final solution that can further trigger protein aggregation. The order in which excipients are added is also important in determining product quality. For example, addition of polysorbate is often performed after any UF/DF step in order to minimize loss due to membrane interaction.
Small-scale characterization studies should be conducted to evaluate protein stability under formulation conditions. Buffers should be characterized to establish appropriate tolerances around excipient concentrations, pH, conductivity and osmolality. Hold time studies using the worst case scenarios for surface area exposed per unit volume should be designed to study the impact of different contact materials on protein stability. Temperature excursions should also be evaluated through hold time studies at various temperatures over prolonged duration of time. The characterization of the mixing process should include both product homogeneity testing as well as impact of mixing shear on product quality. Based on the tank and impeller geometry and product properties (viscosity and density), bulk and impeller tip shear can be computed for the manufacturing conditions. In the absence of appropriate scaled-down mixing systems, rheometers can be utilized in the lab to expose the product to the maximum applicable shear over the recommended duration of the mixing process. Final samples can be analyzed to assess the impact on product quality attributes. These findings can then be verified with fewer runs on the commercial scale to determine the impact during large-scale processing.
Filtration. After the bulk drug substance has been formulated, it goes through sterile filtration. Sterile filtration is usually performed with a 0.22 μm filter to make sure that the bulk is free from viable micro-organisms. An additional in-line filtration step might be incorporated just before filling. Dual filtration prior to filling may also be employed for risk mitigation in the scenario of a filter failure. The protein solution as a result could see multiple filtration steps before being filled as a final dosage form in the drug product container. It is therefore important to evaluate the impact of these filtrations on product quality. Sterile filtration at high trans-membrane pressure could stress the protein while pushing it through the filter pores.
The protein can also selectively bind to the membrane resulting in either misfolding on the membrane surface or protein loss. It is therefore important to study the compatibility of the product to the membrane material. Figure 5 shows the binding of a protein drug product (at 1 mg/mL) to PVDF membrane. It is seen that in this case, up to 37.5 μg of protein is adsorbed per unit cm2 of the filter area. The loss could be appreciable for low concentration products if the batch size is relatively smaller or if the bulk is not being pooled after filtration and before filling.
Similarly other formulation components, e.g., surfactants that are added as stabilizers, can get adsorbed on the membrane surface. This will cause the surfactant concentration in the solution to go below the target, which might result in product destabilization. Recent study by Mahler et al. (40) reported minimal loss of polysorbate 20 due to adsorption on filter membrane and also suggested that protein in formulation could influence surfactant concentration during dialysis process. While such losses may not be significant, it is advisable to test polysorbate concentrations under the final scale process conditions. A larger filter area will reduce filtration time, but at the same time it will maximize the protein and excipient losses associated with membrane adsorption. As a result, scale-down studies should be conducted prior to large-scale processing to assess the impact of filtration on product quality and to recommend the optimum filter size and the membrane type for the manufacturing process. Other filtration process parameters such as the trans-membrane pressure across the membrane, the temperature of the bulk and the liquid flow during filtration should also be evaluated for their impact on protein stability.
Drug Product Filling. Once the drug substance has been formulated and sterile filtered, it is filled into the primary drug product containers, which are usually vials, or syringes for prefilled injectables. During this step, the drug product not only comes in contact with the primary container but also the various components such as stoppers, plungers, etc. All of these components are subjected to sterilization processes separately and brought together under aseptic processing conditions (41). Since there is no further sterilization step, it is critical to maintain the sterility of the drug product during this step. The environment during the filling process could also contribute to foreign contaminants in the final drug product. The container closure systems and the environment of the fill chamber are qualified to be of the highest standards (class 100 room) needed to ensure product quality. Air flow patterns, HEPA filtration, humidity and operation design are used to minimize sources of foreign contaminants such as airborne dust, depyrogenation particles and fibers from operator garments, mobile machine parts and components. In addition to creating additional solid−liquid interfaces that may deactivate proteins, foreign particles pose a significant risk of causing immunogenicity (42, 43).
Interactions with container surface and components, which come in direct contact with the drug product, can also affect protein stability. Siliconized stoppers can contribute to protein aggregation (44) and particulate formation in vials. Leachables and extractables from the container/component surface can further impact the physical and chemical stability of the drug product. Such protein stability issues are also common to primary container and delivery devices and are discussed in detail in the later section “Drug Product Storage, Transport and Delivery”.
During filling the drug product can be subjected to high shear that can cause protein unfolding (45). Biddlecombe et al. (38). have reported significant levels of protein aggregation and precipitation in therapeutic antibodies due to shear in the presence of solid−liquid interfaces. Filing procedure can also affect protein stability. For products sensitive to oxidation and deamidation, fill can be performed under nitrogen to minimize degradation losses (46). The shape and height of the filling nozzle, as well as the fill speed, could contribute to foaming in vials that can possibly result in exposing the hydrophobic residues of the protein to air and initiate protein aggregation at the air/water interface. Fill parameters including the filler speed and nozzle design should be optimized to avoid other issues such as dripping and liquid spurting. For frozen formulations, it is important to have the fill volume in the vials below a threshold level to avoid any vial breakage issues. For lyophilized products, mannitol crystallization is attributed to be the cause of vial breakage during freeze-drying (47). Here too, it is recommended to keep the fill volume within 30% of the volumetric capacity of the vial. The filling speed also governs the process time for the fill step and hence the exposure to ambient temperature. Since filling is often done at room temperature, the stability data of the product should support the total exposure time for each processing step. Exposure to temperatures and durations outside the recommended window could affect product stability. Light exposure during the filling and post-filling processes may also result in protein degradation for certain products. These issues are discussed in detail under the photostabilty section. The other key factors include residual moisture level in stoppers and the permeability of water vapor through the stopper. These are critical for lyophilized formulations and are further discussed in the following section. Other device related issues are discussed in the section on drug product storage, transport and delivery.
Lyophilization. Protein therapeutics can exhibit limited stability toward physical and covalent degradation phenomena. To achieve acceptable shelf life, proteins drug products are often lyophilized (for expanded reviews, see refs 48−50). The process of freeze drying, or lyophilization, removes most of the water in the drug product, except that which is presumably associated with the protein (typically less than 1%). Since the presence of water is required for most covalent degradation phenomena, such as residue fragmentation and isomerization events, the process of lyophilization is usually an effective way of overcoming these instabilities. Further, since diffusion is minimized in the solid compared to the liquid state, phenomena such as aggregation, which require diffusive encounters between monomers, can also be effectively prevented.
However, the process of lyophilization can produce instability in proteins, typically manifested by either an irreversible change in structure or greater levels of aggregation in the lyophilized product (51). There are three main steps in lyophilization, namely, freezing, primary drying and secondary drying (see Figure 6). As the name implies, the purpose of the freezing step is to convert water into ice. The frozen product is thought to typically contain 18−20% water (48) and all of the solutes in a single amorphous phase, other than the ice.
As mentioned in earlier sections, this step is prone to creating a number of potential instabilities. The removal of bulk water from the surrounding of the protein can reduce the magnitude of the hydrophobic effect that keeps protein structures folded. Further, proteins may on occasion denature due to adsorption to ice/water interfaces (25). Several reports have shown that the rate of freezing can play a significant role in protein aggregate formation, both soluble and insoluble (24, 28). Non-ionic surfactants, such as polysorbates, are commonly included in formulations to prevent freeze/thaw-related stresses, presumably by shielding denaturation at the ice−water interface (52).
Another important stress to keep in mind during the freezing step is the formation of the “freeze concentrate”, a network of residual water, excipients and protein that are left behind in a different phase after ice crystallizes (48−50,53). The increase in effective protein concentration can facilitate protein aggregation, while the increase in ionic strength can at times facilitate covalent modifications, especially if there is a change in pH due to crystallization of buffer components. Interestingly, potassium phosphate buffers are preferred over sodium phosphate due to the much smaller change in pH upon freezing (49). A recent study using lactate dehydrogenase as a model showed that no significant degradation occurred when the protein was exposed to the constituents in the freeze concentrate in the absence of ice, suggesting that ice formation during freezing may a key destabilizing element, at least for this particular protein (53).
An annealing step, which is holding the system for a period of time at a temperature between the ice melt and the glass transition temperature of the freeze concentrate, is often performed after freezing (54). There are two purposes behind this step: (a) to crystallize out bulking agents such as glycine and mannitol and (b) to allow larger ice crystals to grow, a process called Ostwald ripening (54). The result of annealing is greater purity of the amorphous phase, which results in greater homogeneity of freeze-dried cakes from one manufacturing run to the next.
During the primary drying step, the temperature is held 2−5 °C below the glass transition of the freeze concentrate, Tg&';, and there is a vacuum applied, which shifts the ice/water/vapor equilibrium in the pressure/temperature phase diagram in such a way that ice is directly converted to water vapor, which is subsequently removed. Holding the shelf temperature too close to the Tg&'; (or above the Tg&';) has the risk of product cake collapse if a subgroup of vials are subjected to greater radiative heat transfer than other vials. On the other hand, if the temperature is held too far below the Tg&'; (>5 °C, for example), it would result in a prolonged primary drying step and thus would not be efficient (55).
Once all the ice has been removed by sublimation, the secondary drying step aims to remove most of the remaining water that did not form ice (as indicated earlier, this could be ∼20%). This dehydration step removes water by desorption at higher temperatures, and so there is a temperature ramp between the primary and secondary drying steps. As the water is removed, the Tg&'; of the amorphous phase increases, but caution must be taken to implement a slow rate of increase in temperature, so as to always stay below the glass transition temperature of the amorphous phase, to prevent cake collapse (55). A large fraction of the remaining water after primary drying may be associated with the protein (“bound water”). Removal of this water may be detrimental to protein structure due to the reduction in stabilizing hydrogen bonds between protein and water molecules. It is believed that the stabilizing effect of saccharide and polyol excipients, such as sucrose, stems from their ability to “substitute” the loss of hydrogen bonding between water and protein with new hydrogen bonds between polyol (-OH group) and protein (56, 57). Monitoring protein secondary structure by Fourier transform infrared spectroscopy (FT-IR) has shown that to preserve secondary structure in the freeze-dried cake, the addition of a stabilizer such as sucrose is essential (58). Recent advances in monitoring protein structure include in situ secondary structure measurements (59) and use of near-IR spectroscopy (60). It is important to note that this stabilizing effect would be lost if the stabilizing excipients were to crystallize out and not remain in the protein-containing amorphous phase. Non-reducing saccharides, such as sucrose and trehalose, typically stay in the same phase as that of the protein and are thereby able to exert their stabilizing effects. This hydrogen bonding between excipient and protein is also thought to be essential to the formation of a robust glassy matrix (49).
The components of the vial (primary container) can play a significant role in the stability of a lyophilized drug product. Special rubber stoppers that allow venting need to be used, so that the sublimed water vapor can exit the vial and be dispersed into the surrounding. The processing history of the stopper can also have an impact on stability; if the stoppers are not adequately dehydrated, then moisture can desorb during storage and potentially destabilize the drug product. Fluoro-elastomer stoppers are recommended to minimize the extent of leaching from the rubber surface. Finally, it is important that the glass bottom be flat; having a convex bottom can hinder efficient heat transfer between the shelf and the vial (42).
In addition to the factors described above, there can be problems encountered during scaling up the lyophilization process, particularly in the areas of heat and mass transfer. Factors that introduce heterogeneity between the vials would need to be optimized during process scale-up. For instance, sublimation rates for vials in the front can be significantly different than for the ones in the center (61). In this case, it was attributable to increased radiation rates in the front, which could be minimized by aluminum foil covering on the inside door. Another factor that can contribute to a success of a large-scale freeze/drying cycle is control of supercooling, which is the difference in the equilibrium ice formation temperature and the actual temperature ice begins to form. In general, reduced supercooling is desirable since it produces larger ice crystals, which minimizes the ice specific area and thereby protein denaturation on ice−protein interfaces (62). A very slow cooling rate may decrease supercooling but may promote phase separation when stabilizing polymers are used in the formulation (63). A moderate supercooling rate in this case would be the best compromise. It was reported that the precooling method the shelf can produce greater vial-to-vial heterogeneity than the slow cooling method (0.5 °C/min) (64). In another study, Rambhatla et al. reported a correlation between the specific surface area of the lyophilized cake and the degree of supercooling and the degree of resistance of the product to mass transfer, which is useful for process design (65). During process scale-up, differences in heat and mass transfer between laboratory-scale, pilot-scale and production-scale lyophilizers must be taken into account. Areas of consideration would include nonuniformity in shelf surface temperatures, the emissivity of various surfaces, resistance of pipes and the performance of the condenser and the refrigeration system. Pikal et al. describe how steady-state heat and mass transfer equations may be used to estimate the effect of lyophilizer variations on the freeze-drying cycle (66). Tang and Pikal provide a detailed review of process parameters that modulate the lyophilization cycle and scale-up (62).
In summary, a selection of appropriate formulation agents and a careful design of the cycle parameters are crucial to the successful generation of a stable, elegant lyophilized protein drug product.
Inspection, Labeling and Packaging. Filled vials are crimped and inspected to ensure the absence of foreign contaminants. Manual or automated inspection can be used at this stage to inspect the filled vials or syringes for presence of visible particles. The labeling and packaging steps, combined with inspection (both manual and automated), would result in exposing the drug product to light for certain duration of time. This photoexposure can result in affecting product quality and should be evaluated a priori for the worst case conditions (maximum intensity and duration). In addition, the interactions between the drug product and primary container during these steps can also impact product integrity. Light sensitivity and packaging related issues are further discussed in the later sections.
Drug Product Storage, Transport and Delivery. After the drug product has been filled in the primary containers, it continues to be in contact with the primary packaging components throughout the duration of storage, transport and delivery. Product-surface interactions with the different contacting materials can have significant impact on the stability and efficacy of protein products (42, 67). Container closure systems could be a major factor in influencing activity and immunogenicity of the product (46). Some of the common factors that can affect the product purity, activity and shelf life during storage, delivery and transport are summarized as follows.
Protein−Silicone Oil Interactions. As mentioned above, protein solutions come in contact with various surfaces during processing and prior to delivery to the patient. Many surfaces, such as syringe, vials, stoppers, and plungers, are treated with different coatings to minimize protein binding and facilitate their processing (such as insertion of stoppers into vials). For instance, rubber stoppers are lubricated with silicone to ensure appropriate closure fitting into the vial neck. Silicone oil is also used in delivery devices to ensure smooth movement of the rubber plunger through the glass barrel. Silicone is considered relatively inert, and its concentration in drug product containers is significantly low to ensure no risk with respect to their biological incompatibility. However, interaction between proteins and silicone oil can induce protein aggregation or product inactivation around the oil droplets (44, 68). While the molecular basis of aggregation induced by silicone oil is not completely understood, it has been reported that no major structural changes are involved (68). Glass surfaces coated with silicone have been reported to increase protein adsorption (69). Alternative coatings in the form of fluoro-elastomer have been introduced for their chemical inertness in addition to the ability to lubricate (42).
Leachables and Extractables. Besides silicone oil, the other constituents of the primary container have the potential to contribute to leachables and extractables, at a concentration that can affect protein stability and activity (32, 42, 70). Leachables are contaminants that have migrated from packaging components (plastic bottles, glass vials, prefilled syringes, rubber stoppers, etc.) or manufacturing equipment (tanks, tubing, etc.) into the product during the drug product processing and over prolonged exposure. Extractables refer to species that are released from the material of construction of the contacting surfaces due to the interaction with product over time under specific solvent and temperature exposure conditions. In addition to toxicological concerns, these chemicals can also impact protein stability. Common leachables and extractables include metal ions, peroxides and aldehydes. The common sources of metal ions are the manufacturing equipment surfaces (e.g., unpassivated stainless steel tanks), glass and plastic packages. Peroxides usually come from packaging or impurities in excipients. Both metal ions and peroxides can induce protein degradation through oxidation via the formation of peroxy radicals (70). Aldehydes on the other hand can react directly with the drug product or excipients and are primarily a concern for low concentration products. Surfactants, which are often used as stabilizers in liquid formulations, may interact with stoppers and affect product quality. Some of the resulting problems associated with liquid prefilled syringes include protein aggregation and in turn carry with them a risk of immunogenicity and active product loss.
Besides the material of the construction, the manufacturing history of the components can play an important role. It has been reported (71, 72) that residual tungsten from the glass syringe manufacturing process can result in aggregation of therapeutic proteins. In this study, the authors reported that the heat used during the molding process contributed to vaporization of the tungsten, which settled in the syringe at the base of the needle. Long-term exposure of the drug product to this residual tungsten resulted in visible particle formation. Such phenomena, unless systematically studied, could be very difficult to resolve during drug product processing. In addition to issues related to protein formulation, mechanical failures in a device can also affect the suitability of drug product for the patient. Needle breakage issues or other problems with plastic parts could result in dosing failure. As a result, prior to the final selection of a delivery device, compatibility studies should be conducted to predict possible effects of the drug product container or device on the stability of drug product and the capability of the device to deliver it. For instance, higher viscosities associated with high concentration formulations may cause challenges while using delivery devices. Finally, stability studies should be designed to monitor product quality over longer times to ensure that the components in contact with the drug product do not have any detrimental effects.
Transportation. Once drug product fill and finish operations are complete, they need to be transported either to various storage sites or to the clinic. Especially for liquid protein drug products, transportation can introduce a substantial risk of causing protein aggregation and particle formation, presumably due to the unfolding of proteins at the air−water interfaces generated by the formation of microbubbles during this process. To minimize the risk at later stages, studies should be performed while screening formulations to ensure resilience to transportation-related stresses. These stresses include those that are related to vibration, shock and changes in pressure and temperature that might be experienced during air or ground transportation. While real-time shipping might adequately capture all these stresses, often this may not be feasible during early formulation development for repeated screening. Various forms of agitation could be used during formulation development, although there is always the question of either under-representing or over-representing transportation related stresses. The United States Pharmacopeia (USP) general chapter 1079 recommends using specifications from the American Society for Testing and Materials document “Standard Practice for Performance Testing of Shipping Containers and Systems” (ASTM D4169−98) and specifications from the International Safe Transit Association (ISTA) for the evaluation of shipping performance of drug products. Vibrational tables are available from various manufacturers to simulate transportation stresses, on which the stress profiles from ISTA or ASTM could be applied. These transportation stresses may result in protein aggregation due to interaction with container surfaces or due to unfolding at air−water interfaces. Surfactants are known to minimize such interactions and are therefore commonly included in the formulation (73). In addition to testing for transportation stresses and addition of formulation components to stabilize proteins against these stresses, adequate testing in the form of visual and subvisible particle assessments and soluble protein aggregation determination should be present. Transportation would not be typically expected to generate covalent modifications in proteins, though testing may be performed for these for added quality assurance.
Photostability. During formulation, fill and finish operations, including visual inspections and packaging, there is exposure of the drug product to various types of light, which may include some degree of UV. Due to the nature and number of unit operations, such as bulk drug substance thaw, filtration, filling into vials and labeling, the therapeutic under consideration may easily be exposed to 2−3 days of ambient light. Light exposure may also affect the stability of the biologic during end user studies, such as dilution and administration via intravenous bags.
In proteins, the residues that are susceptible to modification in the presence of light are tryptophan, tyrosine, phenylalanine and cystine (74). In addition, the peptide backbone is also often a target for photodegradation (74). In the presence of light, peroxy radicals are often formed, especially in combination with dissolved oxygen (75). These peroxy species can covert methionine to methionine sulfoxide (74, 76) and cysteines to various sulfones and acids (74, 76), while tryptophan can be converted to form kyneurenine and other hydroxylated derivatives (74, 77−79). For instance, intense fluorescent light was able to rapidly convert IgG1 Fc methionines to methionine sulfoxide (80). Of all these degraded products, the kyneurenines have the capacity to absorb longer wavelength light and are therefore colored (81). In the case of phenylalanine, photolysis results in formation of the phenyl radical cation that can react with water to form o-, m- and p-tyrosine byproducts (77−79). Other than residue modification, the most common photodegradation pathway is protein aggregation. Often times, photolytic aggregation occurs in part due to crosslinking of tyrosyl radicals, stemming from tyrosine degradation, at the ortho positions of the tryrosine ring (74). The dityrosine moiety has a characteristic fluorescence at 410−415 nm, following excitation at 315 nm (82). Ultimately, many of these degradations could lead to loss of efficacy of the drug product under consideration.
Any change, in this case light-induced, that affects the quality, safety or efficacy of the drug can potentially be deemed unacceptable from the point of view of patient consumption. Therefore, studies should typically be conducted, with appropriate analytical testing to determine the extent of the degradations described above, to determine the extent of light sensitivity for the drug product at hand. The ICH Q1B guideline recommends performing photostability testing in a controlled light chamber with one of two options for light sources:
(a) A single lamp source with wavelengths below 320 nm filtered out, with an output similar to either D65 (international standard for outdoor light) or to ID65 (standard for equivalent window-filtered indoor direct daylight), or
(b) Two lamp sources, one for fluorescent lighting per ISO 10977 guideline and another for UV in the 320−400 nm range
In the “confirmatory study” approach in ICH Q1B, an overall illumination level of at least 1.2 million lux hours is recommended, with integrated near-UV energy of 200 W h/m2 or more. It is likely that biologics, given their nature, may fail these recommended levels in ICH Q1B. However, it is possible that formulation components, such as antioxidants, may slow down the photodegradation profile of the protein therapeutic under consideration. For these reasons, it would be prudent to incorporate photostability studies while deciding between top formulation candidates and testing for manufacturing suitability, prior to final formulation selection. A good place to start would be to identify the spectral outputs of the various light sources in place, and the lengths of exposure time, during the different bioprocessing steps during bulk manufacturing and drug product fill and finish. Additional estimates would need to be made about clinical light exposure during end user administration. Once these are made, along with the wavelengths recommended by ICH Q1B, a time-dependence study would need to be designed to test the resilience of the biologic to light-induced degradation. From this study, recommendations could then be made about the processing times and light exposure conditions during manufacturing and patient delivery.
Drug product quality is governed by several factors during the processing steps associated with formulation, fill and finish operations. Process operating conditions as well as the various surfaces that come in contact with the drug product can have a direct or indirect influence on the integrity of the product. The damage to protein can manifest as a combination of covalent and physical degradations. It is therefore important to characterize the stability of the protein over the range of process parameters. Scale-down studies designed to capture the worst case scenario at large scale provide a useful way of evaluating such issues and developing recommendations for the manufacturing process. Process-scale-dependent issues however can only be evaluated at large scale. In addition to the operation parameters, the manufacturing history of the various components plays an important role in determining product quality. Fill-finish operations need to be conducted in a high quality aseptic environment to minimize the introduction of foreign contaminants. Trace impurities in excipients, material used for thin coatings on device and container surfaces and excessive light exposure can induce reactions that destabilize the protein and alter the purity, activity and efficacy of drug product.
The authors are thankful to Dave Brems, Ed Walls, Wenchang Ji, Erwin Freund, Wei Liu, Bruce Eu, Margaret Ricci and Kapil Gupta for their helpful comments and suggestions towards this manuscript.