Abstract
Saccharides, including sucrose, trehalose, mannitol, and sorbitol, are commonly employed as stabilizers, cryoprotectants, and/or tonicity adjusters in protein formulations. During the thawing of a protein-containing formulated bulk drug substance conducted prior to a drug product (DP) filling operation, a white, crystalline precipitate was observed. In addition, upon thawing, vial breakage was observed for filled DP that had been previously frozen at −40 °C. To investigate the causes of both phenomena, the freeze/thaw behavior of the formulation components was studied. Multiple physical characterization techniques, including differential scanning calorimetry (DSC), electrical resistance measurements, thermomechanical analysis (TMA), and powder X-ray diffraction (PXRD), were utilized to characterize the formulations. The PXRD pattern of precipitate collected from thawed bulk was consistent with that of a mannitol control. An exothermic transition observed by DSC, a sharp increase in electrical resistance detected via resistivity measurements, and the onset of volumetric expansion of the frozen matrix evident in the TMA curve offer evidence that the frozen mannitol solution undergoes transitions at or near the vial breakage temperature (−22 to −23 °C) observed during warming. In addition, osmolality measurements taken from fractionated aliquots of frozen samples indicated that non-uniform concentration gradients contributed to precipitation of mannitol observed at larger scales. Small-scale laboratory experiments (i.e., 10–125 mL) failed to adequately predict behavior at larger scale (i.e., in 1 L and 2 L bottles). Upon linking the detrimental behavior to the freeze/thaw properties of the tonicity adjustor, mannitol, alternative saccharide excipients, including sorbitol, sucrose, and trehalose, were evaluated at isotonic concentrations over a temperature range of −80 to 25 °C using physical-chemical techniques and visual observation. Neither precipitation nor vial breakage was observed for the alternate saccharides. Recommendations for saccharide selection are provided based on storage conditions and scale considerations for liquid biopharmaceutical formulations.
LAY ABSTRACT: Saccharides, including sucrose, trehalose, mannitol, and sorbitol, are commonly employed as stabilizers, cryoprotectants and/or tonicity adjusters in protein formulations. During thawing of formulated bulk drug substance, a white, crystalline precipitate was observed. In addition, upon thawing, vial breakage was observed for filled drug product that had been previously frozen at −40 °C. To investigate the causes of both phenomena, multiple physical characterization techniques were utilized to characterize the formulations. The powder X-ray diffraction pattern of precipitate collected from thawed bulk was consistent with that of a mannitol control. Upon linking the detrimental behavior to the freeze/thaw properties of the tonicity adjustor, mannitol, alternative saccharide excipients, including sorbitol, sucrose, and trehalose, were evaluated at isotonic concentrations over a temperature range of −80 to 25 °C using physico-chemical techniques and visual observation. Neither precipitation nor vial breakage was observed for the alternate saccharides. Recommendations for saccharide selection are given based on storage conditions and scale considerations for liquid biopharmaceutical formulations.
- Mannitol
- Freezing
- Thawing
- Protein formulation
- Vial breakage
- Crystallization
- Saccharides
- Scale-up
- Formulated bulk drug substance
- Tonicity
- Excipient
Introduction
A robust formulation should maintain product quality and physical and chemical stability of the active ingredient under the conditions required for manufacturing, storage, and shipment. Solutions of biopharmaceuticals often undergo freeze/thaw operations at various stages during development and manufacturing. For example, bulk drug substance (BDS) is often frozen to prolong shelf-life, allow operational flexibility, and maintain quality of poorly stable products (1, 2). In addition, during early stages of development when limited stability data are available, frozen conditions may be adopted for storage of toxicology and early clinical supplies, as well as shipment of stability samples to testing laboratories. For these reasons, freeze/thaw studies and characterization of the thermal properties of the active ingredient are conducted as part of routine development activities. However, in addition to the active ingredient, the composition of the formulation typically includes a variety of excipients, the properties of which may vary as a function of freeze/thaw conditions and affect product attributes.
Based on the administration requirements and physical-chemical properties of the active ingredient, a variety of functional excipients may be added for stabilization of the active ingredient, facilitation of clinical administration, and/or maintenance of product integrity (3, 4). Tonicity adjusters are added to injectable solutions to control osmolality and thereby minimize potential for osmotic shock at the site of injection following administration, which can lead to local irritation (5). Typical excipients used for tonicity adjustment include sodium chloride, glycerin, mannitol, dextrose, and trehalose (6). However, because they may compromise physical stability of active ingredients, ionic tonicity adjustors are often avoided for solutions of biopharmaceuticals (7). Therefore, saccharides, which are non-ionic and pharmaceutically acceptable, are commonly employed as tonicity adjusters in injectable formulations (6, 8). In addition, saccharides are one of the most commonly used stabilizers and/or cryoprotectants in both solution and lyophilized biologic formulations (6). The freeze-related properties of commonly used pharmaceutical excipients, including saccharides, and their impact to protein stability have been well characterized for purposes of lyophilization cycle design (8⇓⇓⇓⇓⇓–14). Given the prevalence of saccharide excipients in biopharmaceutical formulations, the increasing emphasis on development of solution formulations requiring incorporation of saccharides into formulated BDS, and the potential need for frozen storage to maximize drug product (DP) and/or BDS shelf-life, it is important to examine the phase behavior of these excipients under freeze/thaw conditions that might be encountered during development, storage, shipping, and manufacturing. The crystallization tendency of saccharide solutes in frozen solutions can affect stability of biopharmaceuticals and is complex and governed by a variety of factors such as compatibility between the formulation components (12, 15).
Small molecule solutes, such as saccharides, are generally quite stable toward freezing because the rates of most chemical reactions are accelerated by elevated temperature according to Arrhenius principles. However, the physical phase behavior of solutes, which is somewhat less predictable, can be dramatically affected by temperature. The most significant effects of temperature on solutes include changes in solubility values and potential crystal form conversion (16). While the relative thermodynamic stability and solubility of crystalline forms in aqueous media are inherent attributes of the given solute, the equilibrium solubility of a given solute form is typically expected to decrease with decreasing temperature. Accordingly, the aqueous solubility of excipients can vary as a function of processing and storage temperature. However, in aqueous solution, the apparent solubility of excipients can also be affected by a phenomenon commonly referred to as cryoconcentration. During freezing of aqueous solutions, pure water freezes to form ice, resulting in exclusion and subsequent concentration of the solute content. This phenomenon, which can lead to concentration of both excipients and active ingredients, is maximized by slow freeze/thaw, which often accompanies increases in scale due to reduced heat transfer rates in larger vessels (17, 18). If the extent of cryoconcentration generated during large-scale freezing exceeds the equilibrium solubility of a solute, including both active ingredients and excipients, precipitation can occur (19). Precipitation of excipients can raise serious quality concerns. For example, Piedmonte et al. demonstrated that crystallization of sorbitol from frozen solutions lead to aggregation of an Fc-fusion protein (20). Aside from concerns related to potential for compromised stability of the active ingredient, the presence of precipitate in bulk solutions would likely not be acceptable for good manufacturing practice processes where it might be considered indicative of a lack of process control. In such cases, the presence of solid particulate might even result in batch rejection even if the precipitate could be dissolved with additional manufacturing steps. Hence, understanding of the freeze/thaw-related phase behavior of formulation components and selection of suitable excipients is requisite for the development and manufacture of robust liquid biopharmaceutical products.
During the drug product manufacturing of a solution formulation, which contains 5% (w/v) mannitol, a precipitate was observed upon thawing of the bulk drug substance, which had been stored at −70 °C and thawed at 5 °C. In addition, vial breakage was observed for DP that had been previously frozen at −40 °C and then thawed at 5 or 25 °C. Both of the BDS and DP (5–6 mg/mL protein) included 5% (w/v) mannitol in 10 mM sodium acetate, pH 5.0. To investigate the causes of and mitigate the precipitation and vial breakage observed in the BDS and DP, the freeze/thaw behavior of the formulation components was studied. Upon linking the detrimental behavior to the freeze/thaw properties of mannitol, the tonicity adjuster, alternative amorphous saccharide excipients including sorbitol, sucrose, and trehalose, were studied over a temperature range of −80 to 25 °C. The crystallization behavior of mannitol and its relevance to lyophilization cycle development have been well studied (8, 11, 13, 21, 22). The current study examines the impact of crystallization and freeze/thaw behavior of mannitol on a formulated BDS and the corresponding filled solution dosage form. Multiple physical characterization techniques, including differential scanning calorimetry (DSC), electrical resistance measurements, thermomechanical analysis (TMA), and powder X-ray diffraction (PXRD), were utilized to characterize the solution formulations selected in this study. Recommendations for saccharide selection are given based on the storage conditions and scale considerations for liquid formulations of biopharmaceuticals.
Materials
BDS was manufactured at a protein concentration of 6 mg/mL at a third-party site under the coordination of the Biologic Third-Party Manufacturing Division of Bristol-Myers Squibb Company (Syracuse, NY), filled into 2 L bottles and then frozen at −70 °C for further use in DP manufacturing operations. DP was prepared by the Clinical Supply Operations Division of Bristol-Myers Squibb Company (New Brunswick, NJ) by thawing the frozen BDS at 5 °C, diluting with formulation buffer to a target concentration 5 mg/mL, and filling (5 mL) into 10 cc Type I flint glass tubing vials which were then stoppered, sealed, and capped. After manufacturing, the DP vials were stored at 5 °C. United States Pharmacepeia/European Pharmacoepeia (USP/EP)-grade mannitol, sucrose, sorbitol, and, trehalose were purchased from J. T. Baker (Phillipsburg, NJ). Type I glass vials were purchased from Alcan Packaging (Syracuse, NY), and fluorotec-coated serum stoppers were purchased from Daikyo Seiko (Tokyo, Japan).
Methods
Effect of Scale on Freeze/Thaw Behavior of the Formulation Components and Different Saccharide Excipients
Various volumes (1.8 L, 0.8 L, 100 mL, and 5 mL) of sterile-filtered BDS—containing 6 mg/mL protein, 5% (w/v) mannitol in 10 mM sodium acetate buffer, pH 5.0), 5% (w/v) mannitol, 5% (w/v) sorbitol, 9.4% (w/v) sucrose, 9.4% (w/v) trehalose in the formulation buffer or the formulation buffer (10 mM sodium acetate, pH 5.0) without mannitol were filled in 2 L, 1 L, 125 mL, and 10 mL sterile polycarbonate bottles. The bottles were frozen at −80 °C (REVCO Ultima II freezer, GS Laboratory Equipment, Asheville, NC) for 48 h and subsequently thawed at 5 °C (Enviromental Growth Chamber, Chagrin Falls, OH) without agitation. The visual appearance of the thawed solutions was recorded. To further investigate the impact of scale on mannitol precipitation, frozen blocks of 5% (w/v) mannitol aqueous solution which had been frozen at −80 °C in 2 L, 125 mL, and 50 mL containers as described above were fragmented regionally using a manual saw to generate aliquots of frozen solution representing different locations within the containers. The aliquots were then thawed at room temperature and analyzed using a vapor pressure osmometer calibrated with OPTI-MOLE osmolality standards at 100, 290, and 1000 mmol/kg (Wescor, Logan, UT). Additionally, 10 cc Type I glass tubing vials were washed with water for injection (WFI), USP, autoclaved, and filled with 5 mL of the above solutions previously sterile-filtered using Millipore Stericup, 0.22 μm polyvinylidene fluoride (PVDF) membrane (Billerica, MA). The filled vials were sealed and placed in a −40 °C temperature chamber (model 115, Test EQUITY LLC, Thousand Oaks, CA) overnight without agitation and then thawed at room temperature. The vials were inspected for breakage during freezing and thawing.
Analysis by Differential Scanning Calorimetry (DSC)
Samples of DP containing 5% (w/v) mannitol, the DP formulation buffer containing 5% (w/v) mannitol, and DP formulation buffer in the absence of mannitol were analyzed using conventional DSC and modulated DSC (mDSC). Conventional DSC traces were collected over a temperature range from −40 to 25 °C using a DSC Q1000 (TA Instruments, New Castle, DE) with a refrigerated cooling accessory. mDSC, an extension of conventional DSC in which sinusoidal modulation of the temperature is superimposed on the underlying heating rate, was utilized to separate and study thermal signals arising from reversible events such as glass transition and non-reversible events such as crystallization (23). Approximately 10 mg of each sample solution was loaded into an aluminum pan (TA Instruments), the sample weights were recorded, and the pans were sealed hermetically. The sample was then cooled from room temperature to −60 °C at a cooling rate of 5 °C/min. The sample was equilibrated at −60 °C for 30 min and then heated to 25 °C at a heating rate of 5 °C/min. The DSC cell was purged with nitrogen as required. For the mDSC mode, the samples were cooled to −60 °C and then heated to 25 °C at the same heating and cooling rate of 1 °C/min. The run was modulated with an amplitude of ±0.5 °C and a period 100 s. Additionally, in order to identify alternative stabilizers/tonicity adjusters for the DP, isotonic samples of 5% (w/v) mannitol, 5% (w/v) sorbitol, 9.4% (w/v) sucrose and 9.4% (w/v) trehalose in the formulation buffer were subjected to the conventional DSC analysis as well using the same system. The formulation buffer in the absence of saccharides was used as a control. For these solutions, the sample was cooled from room temperature to −60 °C at a cooling rate of 5 °C/min. The sample was held at −60 °C for 30 min and then heated to 25 °C at a heating rate of 5 °C/min.
Resistivity Measurements
A VirTis Genesis lyophilizer (VirTis Co., Gardiner, NY) was used to measure the change in resistance of samples as a function of sample temperature. Eight 10 cc, USP Type I flint glass tubing vials filled with 5 mL of 5% (w/v) mannitol solution were placed in the lyophilizer and cooled to −40 °C at a rate of approximately 2 °C/min, the maximum cooling rate of the lyophilizer. The samples were held at −40 °C for 30 min and then heated to 25 °C at a heating rate of 1 °C/min. The temperature and resistance profiles of the samples were monitored simultaneously as described below. The resistance of the samples was measured with a resistance probe made up of two cylindrical stainless steel rods of approximately 1.2 mm in diameter that were spaced about 7 mm apart. Voltage applied by the probe induces a flow of current in the solution for which the resistance is being measured. The temperatures of the samples were monitored using Ni-K thermocouple probes (VirTis Co.). The vial breakage during the heating step was audible and observed visually. Breakage temperatures were recorded.
Thermal-Mechanical Analysis (TMA)
In order to measure the dimensional changes as a function of temperature, TMA measurements were conducted at Mettler Toledo (Materials Characterization Group, Columbus OH) using Mettler Toledo TMA/SDTA 841 Thermomechanical Analyzer. Samples of Milli-Q water, 5% (w/v) mannitol in Milli-Q water, the formulation buffer in the absence of mannitol, and the formulation buffer in the presence of 5% (w/v) mannitol were prepared as described above. Aliquots (100 μL) of the sample were loaded into an aluminum crucible, and the dimensional changes of the samples were measured by a quartz-glass measuring probe. The probe location was zeroed while in contact with the bottom of the aluminum crucible. The sample was cooled from room temperature to −60 °C and held isothermally at −60 °C with a liquid nitrogen cooling accessory for 30 min prior to starting the heating ramp at 1 °C/min. A change in the dimension of the sample on the sample support was recorded as a change in the position, or displacement, of the measuring probe. The TMA warming thermograms were normalized to the original sample thickness at −60 °C.
Powder X-Ray Diffraction (PXRD)
PXRD patterns were collected using a Bruker C2 GADDS (Madison, WI) with Cu K α (40 KV, 40 mA) radiation. The sample detector distance was 15 cm. Powder samples were placed in sealed glass capillary tubes of 1 mm or less in diameter, which were then rotated during data collection. Data were collected for 3 ≤ 2θ ≤ 35° with a sample exposure time of at least 1000 s. The resulting two-dimensional diffraction arcs were integrated to create a traditional one-dimensional PXRD pattern with a step size of 0.02 degrees 2θ in the range of 3 to 35 degrees 2θ.
Results
Macroscopic Observations of Freezing Behavior of Different Saccharide Excipients
Table I summarizes the effects of saccharide species and scale on precipitation observed following freezing at −80 °C and thawing at 5 °C. A significant amount of precipitate formed in the 1 L and 2 L bottles containing the BDS or 5% (w/v) mannitol solution, but no precipitate was evident in the 1 L and 2 L bottles that contained 5% (w/v) sorbitol, 9.4% (w/v) sucrose, 9.4% (w/v) trehalose solution, or the formulation buffer without mannitol. In contrast, no precipitate formed in the 10 mL and 125 mL bottles containing any of the solutions studied.
To further investigate the impact of scale on mannitol precipitation, blocks of 5% (w/v) mannitol solution frozen at −80 °C in 2 L, 125 mL, and 50 mL containers were fragmented regionally, thawed, and analyzed by osmometry. The osmotic concentrations of samples from the top, middle, and bottom of the 125 mL and 50 mL frozen blocks ranged from 270 to 384 mOsm/kg. However, the osmotic concentrations of samples collected from various locations of the 2 L frozen block varied over an order of magnitude from 178 to 1295 mOsm/kg (Figure 1). Precipitate was observed and higher osmotic concentrations were detected in positions closest to the center of the block. In fact, samples from the center positions of the 2 L block contained saturated solutions in which precipitate could only be partially dissolved. In these samples, the osmotic concentration of the supernatant was 1295 mOsm/kg. As controls, the osmolality of 5%, 10%, and 15% (w/v) aqueous mannitol solutions was measured at 287, 603, and 872 mOsm/kg, respectively, while that of the supernatant from an 18% (w/v) saturated aqueous mannitol solution was 1472 mOsm/kg.
Table II shows the extent of vial breakage when the vials were subjected to the temperature cycle of freezing at −40 °C followed by warming at room temperature. Twelve out of 20 (60%) of the vials containing DP and 15 out of 20 (75%) vials filled with 5% (w/v) mannitol solution broke during the warming phase of the cycle. However, each of 20 vials filled with 5% (w/v) sorbitol, 9.4% (w/v) sucrose, 9.4% (w/v) trehalose solution, or the formulation buffer in the absence of mannitol were intact following the freeze/thaw cycle. No vial breakage was observed during the freezing phase for any compositions evaluated.
Identification of the Precipitate in the 1 L and 2 L Bulk Solutions
PXRD was used to determine the identity of the precipitate in the BDS and mannitol bulk solutions. The pattern of the collected precipitate (Figure 2) is consistent with that obtained from mannitol granular. Differences in peak intensities are due to preferred orientation caused by morphology and particle size differences between the two samples (data not shown).
Thermal Characteristics of the Drug Product and the Formulation Buffer
Due to sample deformation and phase transformation (i.e., liquid to solid state) upon freezing, it was not possible to analyze formulations by DSC during the freezing step. However, the thermal transitions could be readily observed from the heating curves. Figure 3a shows the DSC heating thermograms of the DP fill (which contains, among other components, 5% w/v mannitol), the DP formulation buffer (without the active species), and the DP formulation buffer in the absence of mannitol. The DSC heating thermograms for all of the samples contain a strong, broad endotherm at about −1 °C (Tpeak) which is attributed to the melting of ice. In the DP sample, an active-related endothermic transition is evident at about −13 °C. Comparison of the heating thermograms reveals the presence of an exothermic transition at about −23 °C in the samples of the DP fill material and the DP formulation buffer containing 5% (w/v) mannitol. However, this event was not observed for the sample of the DP formulation buffer without mannitol. Because the sloping baseline leading into the exothermic transition suggested the possibility of coupled transitions, the mannitol-containing samples were analyzed using mDSC. The mDSC results (Figure 3b) illustrate that the thermal event at about −23 °C consisted of coupled reversible and non-reversible transitions, consistent with a glass transition event which is immediately followed by crystallization.
Figure 4 shows DSC heating thermograms of 5% (w/v) mannitol, 5% (w/v) sorbitol, 9.4% (w/v) sucrose, 9.4% (w/v) trehalose in the formulation buffer, and the formulation buffer alone in the absence of these saccharides. The heating curve of 5% (w/v) mannitol solution contains an exotherm at about −23 °C, but the heating curves of the other four solutions do not show the exotherm at about −23 °C. In all samples, the DSC heating curves contain only one endothermic transition between 0 and about −1 °C, which is attributed to the melting of crystalline ice.
Correlation of Vial Breakage with Electrical Characteristics
Figure 5 shows the resistance-temperature profile of a 5% (w/v) mannitol solution during warming following cooling at −40 °C. Resistivity in the frozen solution decreased with increasing temperature up to −23 °C, and then increased sharply over a temperature range of 1 °C to a maximum at −22.7 °C. It then decreased again gradually with increasing temperature until −13 °C. Table III contains the observed breakage temperatures of the eight vials containing 5 mL of 5% (w/v) mannitol solution, monitored simultaneously with the resistance. Vial breakage occurred over a temperature range that is associated with a sharp increase in the resistance of the frozen matrix.
Thermal-Mechanical Analysis (TMA)
Figure 6 shows TMA thermograms of Milli-Q water, 5% (w/v) mannitol in Milli-Q water, the DP formulation buffer in the absence of mannitol, and the formulation buffer in the presence of 5% (w/v) mannitol. The TMA thermograms were normalized to the original sample thickness at −60 °C. The Milli-Q water and the formulation buffer without mannitol showed straight lines except for a sharp decrease at about 0 °C, corresponding with the melting point of the bulk ice. In contrast, 5% (w/v) mannitol in Milli-Q water and the formulation buffer in the presence of 5% (w/v) mannitol showed a transition at about −23 °C, which represents a sudden increase in dimensional volume of the sample.
Correlation of Thermal, Electrical, Mechanical Analysis, and Vial Breakage Temperature
Figure 7 summarizes the DSC, resistivity, TMA, and vial breakage results (Figures 3a, 5, 6, and Table III). The overlayed results clearly demonstrate the close correlation among the sharp increase in the DSC exothermic transition, the electrical resistance, the sudden increased dimensional volume of the sample, and the vial breakage temperature. All these events occurred at about −23 °C.
Discussion
Freeze/thaw operations represent a critical step in protein DS and DP manufacturing. Maintaining the protein quality during freezing, frozen storage, and thawing operations is challenging, especially at large scales. During the development and manufacture of a solution formulation, which contains 5% (w/v) mannitol, precipitation was observed upon thawing 2 L bottles containing 1.8 L of BDS at 5 °C. PXRD revealed that the pattern of the precipitate is consistent with that obtained from granular mannitol (Figure 2) and therefore confirmed the identity of the precipitate in the bulk solution. Although significant precipitation was observed in 1 L and 2 L bottles containing 5% (w/v) mannitol solution, it was not observed in the 10 mL and 125 mL bottles (Table I). Furthermore, no precipitation was observed in the 5% (w/v) sorbitol, 9.4% (w/v) sucrose, or 9.4% (w/v) trehalose solution at any the scales studied (Table I).
Cryoconcentration (i.e., when an aqueous solution freezes, pure water freezes first which then concentrates the remaining solute content) is a common phenomenon which can occur during bulk freeze/thaw, and the extent of cryoconcentration is maximized if the rate of freezing is slow (1, 17). The solute distribution that results from following freezing operations can affect protein quality (24⇓⇓⇓⇓–29). Aside from the concern of protein quality/integrity, if the cryoconcentration created during slow freeze/thaw operations is in excess of the solubility of a solute, precipitation will occur. Haripada Maity et al. reported freezing profiles of a solution at −80 °C in a 2 L bottle and two 50 mL tubes monitored by multiple temperature probes (30). It took approximately 12 h to reach a stable temperature of −80 °C throughout the 2 L bottle, yet only 2 h to reach a stable temperature of −80 °C in the 50 mL tubes. Furthermore, the difference of excipient concentrations in the solution between the top and the bottom of the 2 L bottle could be as high as 10 fold following freezing at −80 °C and thawing at 4 °C without mixing. However, the authors found that regional freeze/thaw behavior in the 50 mL tubes is considerably uniform. Similar results were reported in other studies on solute distribution in various freezing systems from 2 L to 20 L containers (24, 26, 31).
In the present study, the osmotic concentrations of samples from the top, middle-center, middle-edge, and bottom of the 125 mL and 50 mL frozen blocks ranged from 270 to 384 mOsm/kg, while the osmotic concentrations of samples collected from various locations of the 2 L frozen block varied over an order of magnitude from 178 to 1295 mOsm/kg. Precipitate was observed and higher osmotic concentrations detected in positions closest to the center of the block. In fact, samples from the center positions of the 2 L block contained saturated solutions in which precipitate could only be partially dissolved. In these samples, the osmotic concentration of the supernatant was 1295 mOsm/kg. The osmolality of 5%, 10%, and 15% (w/v) aqueous mannitol solutions was measured as 287, 603, and 872 mOsm/kg, respectively, while that of the supernatant from an 18% (w/v) saturated aqueous mannitol solution was 1472 mOsm/kg. These data indicate substantial concentration gradient within the 2 L block, throughout which the osmotic concentration varied widely from that of the target mannitol concentration, that is, 5% (w/v). These results demonstrate that scale-related slow freeze/thaw resulted in cryoconcentration of the solute manitol at 2 L scale.
The equilibrium solubilities of mannitol, sorbitol, sucrose, and trehalose are 18%, 220%, 200%, and 69% (w/v), respectively, and the isotonic concentrations of mannitol, sorbitol, sucrose, and trehalose solutions are 5%, 5%, 9.4%, and 9.4% (w/v), respectively (32). Consequently, the solubility of mannitol is low relative to sorbitol, sucrose, and trehalose. In our study, the observed precipitation in the 1 L and 2 L mannitol-containing solutions is consistent with the cryoconcentration phenomenon (i.e., slow rate of heat transfer). The relatively low solubility of mannitol and cryoconcentration formed during freeze/thaw might be responsible for the precipitation observed upon thawing of the DS. Moreover, the available literature suggests that freezing is more rapid and, likely, more uniform throughout the 10 mL and 125 mL bottles, which minimizes the potential for precipitation. Based purely on scale, cryoconcentration might also be expected in the other 1 L and 2 L scale solutions (i.e., sorbitol, sucrose, and trehalose) studied. However, precipitation was not observed at any scale under the conditions studied likely due to the relatively higher solubility of the alternate saccharides evaluated. For DP manufacturing, for example, large polycarbonate carboys (e.g., 10 L and 20 L), are commonly used to freeze and transport bulk drug substance. Because the path lengths are large and the heat transfer is slow, convective effects become important at scale and might affect the distribution of solutes in practical systems (1, 25). As a result, cryoconcentration and excipient solubility become important factors governing product quality in large-scale manufacturing process. Appropriate mixing during thaw is a solution to minimize cryoconcentration-related effects; in addition, excipient solubility should be considered as well. For the BDS and DP involved in this study, the precipitation of mannitol has no impact on the protein stability. However, because solution formulations may potentially undergo bulk freeze/thaw processes, formulations should be optimized to ensure not only the stability of the active ingredient, but the product manufacturability and quality. The need to examine product stability over a broad range of process parameters has been reported in the literature (33). Such an examination might be achieved by characterization studies at small scale using qualified scale-down models or large-scale experiments designed to examine worst-case scenarios related to change in operating conditions (33).
In addition to the precipitation observed upon thawing of the BDS, vial breakage was encountered during warming of the frozen DP solution. As indicated in the results (Table II), vial breakage was observed upon thawing of the frozen DP and the formulation buffer, both of which contain 5% (w/v) mannitol, but not to frozen formulation buffer in the absence of mannitol. In addition, vial breakage was not observed during warming of isotonic sorbitol, sucrose, and trehalose frozen solutions. Accordingly, the vial breakage of the tubing vials used in this study was attributed to the presence of mannitol in the formulation, consistent with the well documented phenomenon of vial breakage by mannitol solutions (21, 34, 35). During lyophilization of mannitol solutions, vial breakage may be mitigated by cooling very slowly, which allows equilibrium crystallization of the water in the solution prior to warming the product (21, 34). However, depending on the available manufacturing infrastructure and site capabilities, controlled-rate cooling (e.g., annealing) is oftentimes not practical or cost-effective during manufacturing-scale freeze/thaw operations, hence freeze/thaw-induced vial breakage may result in product loss and safety issues. Therefore, investigational studies were conducted to better understand the vial breakage phenomenon and support mitigation strategies, in other words, to enable selection of alternative saccharides for use as stabilizers/tonicity adjusters to mitigate vial breakage during freeze/thaw solution formulations.
Comparison of the DSC heating curves (Figure 3a) shows that the exothermic transition at about −23 °C was detected only in the samples containing 5% (w/v) mannitol, but was absent with the samples without mannitol. The thermal characteristics of mannitol have been extensively investigated (8, 21, 23, 34). The exotherm at about −23 °C has been interpreted as crystallization of amorphous mannitol, which occurs during warming a previously frozen solution. In the DP sample (A in Figure 3a), an active-related endothermic transtion is evident at about −13 °C. This event did not affect the exotherm at about −23 °C in the formulation buffer sample (B in Figure 3a), suggesting there is no impact of the active (protein) on the crystallization behavior of mannitol and thereby was not studied further. The mDSC results (Figure 3b) indicate that the thermal event at about −23 °C contains coupled reversible and non-reversible transitions, suggesting that the transition event(s) at about −23 °C in the samples may be attributed to mannitol crystallization and glass transition event of amorphous mannitol preceding crystallization.
As expected, the electrical resistance of a frozen solution normally decreases with increases in temperature. In our study, the resistance of the frozen mannitol solution decreased with increases in temperature to −23 °C; however, it increased over a temperature range of 1 °C to a maximum at −22.7 °C (Figure 5). Interestingly, vial breakage occurred over the temperature range from −25.5 to −23 °C (Table III), which is associated with the sharp increase in the resistance of the frozen matrix.
It is well known that a glass vial filled to capacity with water may break during freezing owing to volumetric expansion of the crystalline matrix of water (i.e., ice formation) upon freezing. In the current frozen mannitol solution study (5 mL in 10 cc glass vials), however, breakage of the vials was observed during warming, not freezing. Because it was postulated that the breakage during warming of the frozen vials must involve expansion of the frozen matrix, TMA was employed to detect the expansion of frozen solutions during warming (Figure 6). The TMA thermograms of water and the formulation buffer in the absence of mannitol were characterized by constant slope until temperatures approaching the melting point of the bulk solution (i.e., ice) at about 0 °C, at which point the melting of crystalline ice leads to a sharp decrease in sample height. In contrast, the TMA thermograms of 5% (w/v) mannitol in water and the formulation buffer in the presence of 5% (w/v) mannitol showed a transition at about −23 °C, which represents a sudden increase in dimensional volume of the sample. These observations are consistent with literature reports (21, 34⇓⇓–37) and offer direct evidence that vial breakage in frozen mannitol solutions is caused by a rapid expansion of the frozen matrix, which occurs at −23 °C, during warming.
Figure 7 summarizes the results of DSC (thermal), resistivity (electrical), TMA (mechanical), and vial breakage temperature from this study. These combined results illustrate clearly that the exothermic transition detected by DSC, the sharp increase in electrical resistance, the volume expansion of the frozen matrix, and vial breakage coincided at about −23 °C. These three techniques offer evidence that the frozen mannitol solution undergoes transitions at or near the vial breakage temperature (−22 to −23 °C) during warming. It can therefore be concluded that a solid phase transition of mannitol, which affects the thermal, electrical, and mechanical properties of the frozen solution, leads to vial breakage in the filled DP.
Saccharides are commonly employed as stabilizers, cryoprotectants and/or tonicity adjusters in protein formulations (6, 8). Therefore, along with mannitol, we also examined the thermal characteristics of sucrose, trehalose, and sorbitol. Figure 4 shows DSC heating curve of 5% (w/v) mannitol, 5% (w/v) sorbitol, 9.4% (w/v) sucrose, and 9.4% (w/v) trehalose in the formulation buffer and the formulation buffer alone in the absence of these saccharides. The heating curve of 5% (w/v) mannitol solution contain an exotherm at about −23 °C, but the heating curves of the other four solutions do not. These results suggest that sorbitol, sucrose, and trehalose do not undergo crystallization, consistent with observations in the lyophilization literature (8, 34). Hence, these alternate saccharides are less likely to cause vial breakage during solution freeze/thaw. This hypothesis is in agreement with the observation given in Table II, where sorbitol, sucrose, and trehalose solutions did not cause any vial breakage while mannitol solution caused substantial vial breakage under the same condition.
Of the commonly utilized saccharide excipients, mannitol is relatively well-studied given its popularity as a bulking agent for lyophilized formulations. For purposes of lyophilization, mannitol possesses many attractive qualities with respect to crystallinity, high eutetic temperature, and matrix properties (8). However, mannitol at or above certain concentrations and volume in glass vials is well-known to cause vial breakage given the propensity of mannitol to crystallize under frozen conditions (21, 34⇓⇓–37). In fact, the freezing behavior of mannitol is well-documented in the lyophilization literature, and preventive measures (e.g., introduction of an appropriate annealing step) are often employed to minimize the potential for vial breakage during processing (21, 34, 35). In addition, mannitol might be utilized as a tonicity adjuster and stabilizer in solution formulations of biopharmaceuticals, for example, prefilled syringes and formulated BDS. However, those solution formulations, as drug product or formulated BDS, often require frozen conditions for storage and global shipment, which may compromise inherent instability of proteins. Controlled-rate cooling (i.e., annealing) is oftentimes not practical or cost-effective during manufacturing-scale freeze/thaw operations. Consequently, amorphous saccharides (e.g., sorbitol, trehalose, and sucrose) are the preferred tonicity adjusters/stabilizers in solution formulations to protect the product against freeze-thaw stresses encountered during transport and storage (6).
Conclusions
The relatively low solubility of mannitol and cryoconcentration resulting from freeze/thaw operations might be responsible for the precipitation observed upon thawing of the BDS. The phase behavior of mannitol affects the thermal, electrical, and mechanical properties of the frozen solution, resulting in vial breakage of the DP. DSC (thermal), resistivity (electrical), and TMA (mechanical) measurements may be used as complementary tools to investigate the mechanism behind vial breakage and test formulations which might cause vial breakage during freeze/thaw operations. The other saccharides (i.e., sorbitol, sucrose, and trehalose) did not cause the observed events. To mitigate the potential for both precipitation and vial breakage, the solubility and crystallization behavior of candidate saccharides for liquid formulations should be characterized and considered in order to ensure integrity of both formulated BDS and DP, and these findings should be considered during process development for both formulation BDS and DP. In addition, from a practical perspective, the findings suggest that, when mannitol is required to impart stability, tonicity, etc. to the drug product, it might be compounded into the solution at the step of the DP fill/finish operations to avoid freeze/thaw induced precipitation in the bulk DS.
Conflict of Interest Declaration
We declare that we have no competing interests.
Acknowledgments
The authors thank Dr. Satish K Singh (Pfizer) for helpful discussions on cryoconcentration; Mettler-Toledo for the TMA support; Ms. Jacqueline Toth (BMS) for generating the PXRD patterns; Mr. Edwin McCloskey (BMS) for assistance with the fractionation; and Dr. Sachin Chandran (BMS) for helping in preparing the figures.
- © PDA, Inc. 2012
References
PDA members receive access to all articles published in the current year and previous volume year. Institutional subscribers received access to all content. Log in below to receive access to this article if you are either of these.
If you are neither or you are a PDA member trying to access an article outside of your membership license, then you must purchase access to this article (below). If you do not have a username or password for JPST, you will be required to create an account prior to purchasing.
Full issue PDFs are for PDA members only.
Note to pda.org users
The PDA and PDA bookstore websites (www.pda.org and www.pda.org/bookstore) are separate websites from the PDA JPST website. When you first join PDA, your initial UserID and Password are sent to HighWirePress to create your PDA JPST account. Subsequent UserrID and Password changes required at the PDA websites will not pass on to PDA JPST and vice versa. If you forget your PDA JPST UserID and/or Password, you can request help to retrieve UserID and reset Password below.