Surface acidity and solid-state compatibility of excipients with an acid-sensitive API: case study of atorvastatin calcium.

The objectives of this study were to measure the apparent surface acidity of common excipients and to correlate the acidity with the chemical stability of an acid-sensitive active pharmaceutical ingredient (API) in binary API-excipient powder mixtures. The acidity of 26 solid excipients was determined by two methods, (i) by measuring the pH of their suspensions or solutions and (ii) the pH equivalent (pHeq) measured via ionization of probe molecules deposited on the surface of the excipients. The chemical stability of an API, atorvastatin calcium (AC), in mixtures with the excipients was evaluated by monitoring the appearance of an acid-induced degradant, atorvastatin lactone, under accelerated storage conditions. The extent of lactone formation in AC-excipient mixtures was presented as a function of either solution/suspension pH or pHeq. No lactone formation was observed in mixtures with excipients having pHeq > 6, while the lactone levels were pronounced (> 0.6% after 6 weeks at 50°C/20% RH) with excipients exhibiting pHeq < 3. The three pHeq regions (> 6, 3-6, and < 3) were consistent with the reported solution pH-stability profile of AC. In contrast to the pHeq scale, lactone formation did not show any clear trend when plotted as a function of the suspension/solution pH. Two mechanisms to explain the discrepancy between the suspension/solution pH and the chemical stability data were discussed. Acidic excipients, which are expected to be incompatible with an acid-sensitive API, were identified based on pHeq measurements. The incompatibility prediction was confirmed in the chemical stability tests using AC as an example of an acid-sensitive API.

Glassy dynamics of sorbitol solutions at terahertz frequencies.

The absorption spectra of D-sorbitol and a range of its concentrated aqueous solutions were studied by terahertz spectroscopy over the temperature interval of 80 K < T < 310 K. It is shown that the slow-down of molecules at around the glass transition temperature, Tg, dramatically influences the thermal dependence of the absorption at terahertz frequencies. Furthermore, two different absorption regimes are revealed below Tg: at temperatures well below Tg, the absorption resembles the coupling of terahertz radiation to the vibrational density of states (VDOS); above these temperatures, between 160 K and Tg, in the sample of pure sorbitol and the sample of a solution of 70 wt% sorbitol in water, another type of absorption is observed at terahertz frequencies. Several possibilities of the physical origin of this absorption are discussed and based on the experimental data this process is tentatively assigned to the Johari-Goldstein ß-relaxation processes shifting to lower frequencies at temperatures below Tg leaving behind a spectrum largely dominated by losses into the VDOS.

Investigation of design space for freeze-drying: use of modeling for primary drying segment of a freeze-drying cycle.

In this work, we explore the idea of using mathematical models to build design space for the primary drying portion of freeze-drying process. We start by defining design space for freeze-drying, followed by defining critical quality attributes and critical process parameters. Then using mathematical model, we build an insilico design space. Input parameters to the model (heat transfer coefficient and mass transfer resistance) were obtained from separate experimental runs. Two lyophilization runs are conducted to verify the model predictions. This confirmation of the model predictions with experimental results added to the confidence in the insilico design space. This simple step-by-step approach allowed us to minimize the number of experimental runs (preliminary runs to calculate heat transfer coefficient and mass transfer resistance plus two additional experimental runs to verify model predictions) required to define the design space. The established design space can then be used to understand the influence of critical process parameters on the critical quality attributes for all future cycles.

Synchrotron X-ray diffraction investigation of the anomalous behavior of ice during freezing of aqueous systems.

Simple aqueous systems, i.e., phosphate-glycine buffers and pure water, were studied at subambient temperatures by X-ray difractometry using a high-intensity synchrotron radiation source at the Advanced Photon Source of Argonne National Laboratory. Complex X-ray diffraction (XRD) patterns, with two or more poorly resolved peaks in place of each of the four diagnostic peaks of hexagonal ice, 100, 002, 101, and 102, referred as "splitting", were observed in the majority of cases. The splitting of up to 0.05 A (d-spacing) was detected for 100, 002, and 101 peaks, whereas 102 peak was less affected. Deformation of the lattice of hexagonal ice, probably due to local stress created on the ice/ice or ice/container interface during water-to-ice transformation, is proposed as a possible mechanism for the observed splitting of XRD peaks. Using molecular modeling, it was estimated that the observed shifts in the peak positions are equivalent to applying a hydrostatic pressure of 2-3 kbars. The splitting can be used to quantify stresses during freezing, which could improve our understanding of the role of water-to-ice transformation on the destabilization of proteins and other biological systems.

Phase transitions in frozen systems and during freeze-drying: quantification using synchrotron X-ray diffractometry.

PURPOSE: (1) To develop a synchrotron X-ray diffraction (SXRD) method to monitor phase transitions during the entire freeze-drying cycle. Aqueous sodium phosphate buffered glycine solutions with initial glycine to buffer molar ratios of 1:3 (17:50 mM), 1:1 (50 mM) and 3:1 were utilized as model systems. (2) To investigate the effect of initial solute concentration on the crystallization of glycine and phosphate buffer salt during lyophilization. METHODS: Phosphate buffered glycine solutions were placed in a custom-designed sample cell for freeze-drying. The sample cell, covered with a stainless steel dome with a beryllium window, was placed on a stage capable of controlled cooling and vacuum drying. The samples were cooled to -50 degrees C and annealed at -20 degrees C. They underwent primary drying at -25 degrees C under vacuum until ice sublimation was complete and secondary drying from 0 to 25 degrees C. At different stages of the freeze-drying cycle, the samples were periodically exposed to synchrotron X-ray radiation. An image plate detector was used to obtain time-resolved two-dimensional SXRD patterns. The ice, beta-glycine and DHPD phases were identified based on their unique X-ray peaks. RESULTS: When the solutions were cooled and annealed, ice formation was followed by crystallization of disodium hydrogen phosphate dodecahydrate (DHPD). In the primary drying stage, a significant increase in DHPD crystallization followed by incomplete dehydration to amorphous disodium hydrogen phosphate was evident. Complete dehydration of DHPD occurred during secondary drying. Glycine crystallization was inhibited throughout freeze-drying when the initial buffer concentration (1:3 glycine to buffer) was higher than that of glycine. CONCLUSION: A high-intensity X-ray diffraction method was developed to monitor the phase transitions during the entire freeze-drying cycle. The high sensitivity of SXRD allowed us to monitor all the crystalline phases simultaneously. While DHPD crystallizes in frozen solution, it dehydrates incompletely during primary drying and completely during secondary drying. The impact of initial solute concentration on the phase composition during the entire freeze-drying cycle was quantified.

Correlation between chemical reactivity and the Hammett acidity function in amorphous solids using inversion of sucrose as a model reaction.

The goal was to evaluate the effects of acidity, expressed as the Hammett acidity function, on chemical reactivity in freeze-dried materials (lyophiles). Dextran-sucrose-citrate and polyvinyl pyrrolidone (PVP)-sucrose-citrate aqueous solutions, adjusted to pH values of 2.6, 2.8, and 3.0 were freeze dried, and characterized by X-ray powder diffractometry, DSC, isothermal microcalorimetry, and Karl Fischer titrimetry. Lyophiles were also prepared from identical solutions but containing bromophenol blue (BB). Diffuse reflectance-visible spectroscopy was used to measure the extent of BB protonation from which the Hammett acidity functions were determined. The stability studies were performed at 60 degrees C. All the freeze-dried samples were observed to be X-ray amorphous with <0.15% w/w water content. The T(g) of dextran lyophiles were approximately 20 degrees C higher than that of PVP lyophiles whereas enthalpy relaxation rates at 60 degrees C were similar. The Hammett acidity functions were significantly lower (i.e., higher acidity) for dextran systems (<2.2-2.6) when compared with PVP systems (3.3-3.9). The rate of sucrose inversion was significantly (an order of magnitude) higher in dextran lyophiles. This study showed that in amorphous matrices with comparable water content and structural relaxation times, chemical reactivity could be significantly different depending on the matrix "acidity".

Comparative rates of freeze-drying for lactose and sucrose solutions as measured by photographic recording, product temperature, and heat flux transducer.

Sublimation from lactose and sucrose solutions has been monitored by temperature measurement, visual observation, heat flux sensing and manometric measurements. Estimates of energy transfer rates to the subliming mass made from visual observations and heat flux measurements are in broad agreement, demonstrating for the first time that heat flux sensors can be used to monitor the progress of lyophilization in individual vials with low sample volumes. Furthermore, it is shown that under identical lyophilization conditions the initial rate of drying for lactose solutions is low with little water sublimation for up to 150 minutes, which contrasts markedly with the much faster initial rate of drying for sucrose solutions. Measurement of the initial heat flux between shelf and vial indicated a lower flux to a 10% lactose solution than to a 10% sucrose solution.

Impact of freeze-drying on ionization of sulfonephthalein probe molecules in trehalose-citrate systems.

"pH memory," i.e., correlation between pH of solution before freeze-drying and chemical reactivity in the freeze-dried state, has been reported in many systems. In this study, the "pH memory" is explored by comparing the extent of protonation of sulfonephthalein probe molecules, bromophenol blue, bromocresol green, and chlorophenol red, in aqueous solution in the pH range of 3.4-6.0 and in the resulting freeze-dried amorphous matrix (lyophile) containing trehalose and sodium citrate buffer. The protonation of the probe molecules was measured in the lyophiles by diffuse reflectance visible spectroscopy, and compared with that in the solution before drying. The protonation of the indicators in the amorphous matrix correlated with solution pH, that is, an increase in solution pH resulted in a progressive decrease in the indicator protonation in the corresponding lyophile. However, the protonation was consistently higher in the lyophile than in the corresponding solution. The Hammett acidity function of lyophiles was calculated based on the extent of protonation of the probe molecules. Protonation of the probe molecules and the Hammett acidity function depended not only on prelyophilization solution pH, but also on the residual water content and the presence of amorphous sugar in the lyophile.

The influence of measurement conditions on the Hammett acidity function of solid pharmaceutical excipients.

In this work the Hammett acidity function has been measured to assess the relative acidity of excipients used in the preparation of pharmaceutical solid dosage forms. A systematic series of experiments is reported which illustrates how the selection of the measurement conditions can influence the results of such determinations. Although the technique is somewhat empirical and relies on several key assumptions it is shown that very consistent results can be achieved by carefully controlling the measurement conditions. It is also shown that by taking this approach laboratory-to-laboratory variation can be reduced to a negligible level and the influences of subtle changes in the acidity of pharmaceutical excipients due to intrinsic variations in their physical properties or due to different processing histories can be detected and quantified.

Thermodynamic and dynamic factors involved in the stability of native protein structure in amorphous solids in relation to levels of hydration.

The internal, dynamical fluctuations of protein molecules exhibit many of the features typical of polymeric and bulk small molecule glass forming systems. The response of a protein's internal molecular mobility to temperature changes is similar to that of other amorphous systems, in that different types of motions freeze out at different temperatures, suggesting they exhibit the alpha-beta-modes of motion typical of polymeric glass formers. These modes of motion are attributed to the dynamic regimes that afford proteins the flexibility for function but that also develop into the large-scale collective motions that lead to unfolding. The protein dynamical transition, T(d), which has the same meaning as the T(g) value of other amorphous systems, is attributed to the temperature where protein activity is lost and the unfolding process is inhibited. This review describes how modulation of T(d) by hydration and lyoprotectants can determine the stability of protein molecules that have been processed as bulk, amorphous materials. It also examines the thermodynamic, dynamic, and molecular factors involved in stabilizing folded proteins, and the effects typical pharmaceutical processes can have on native protein structure in going from the solution state to the solid state.

Partially crystalline systems in lyophilization: II. Withstanding collapse at high primary drying temperatures and impact on protein activity recovery.

In an accompanying article we have described the construction of the water-rich sections of raffinose-glycine-water and trehalose-glycine-water state diagrams. In this study, we use the information obtained from the state diagrams to identify the minimum weight fraction of the crystalline component in glycine-carbohydrate systems necessary to withstand collapse at high primary drying temperatures. We also determine the impact of primary drying, substantially above T'g, on the recovery of lactate dehydrogenase (LDH) activity. Ambient and variable temperature X-ray powder diffractometry and differential scanning calorimetry were used to characterize the frozen and freeze-dried systems. Aqueous solutions with glycine to carbohydrate (raffinose pentahydrate or trehalose dihydrate) weight ratios ranging from 0.2 to 2.0 were freeze dried. The protein formulations contained 20 mM citrate buffer (pH 6.0) and LDH (20 microg/mL). A glycine to anhydrous raffinose weight ratio >or=1.18 and a glycine to anhydrous trehalose weight ratio >or=1.56 were necessary to withstand macroscopic collapse in the system, when the primary drying was carried out at a product temperature at least 10 degrees C above the T'g. The recovery of LDH activity was almost complete in the reconstituted lyophile whether the primary drying was carried out above T'g (-10 degrees C) or below T'g (-32 degrees C). Thus, by judiciously combining crystalline and amorphous components, it was possible to primary dry at temperatures substantially above the T'g.

Partially crystalline systems in lyophilization: I. Use of ternary state diagrams to determine extent of crystallization of bulking agent.

Two model ternary systems: water-glycine-raffinose and water-glycine-trehalose were investigated to determine the extent of glycine crystallization in frozen solutions. The use of such partially crystalline systems allows primary drying to be carried out substantially above the collapse temperature. Differential scanning calorimetry (DSC) and variable temperature X-ray diffractometry (XRD) were used to monitor phase transitions in frozen systems as well as to determine the T'g. Aqueous solutions containing different glycine to carbohydrate weight ratios were first cooled to -60 degrees C and then warmed to room temperature. In both raffinose and trehalose systems, when the initial glycine to sugar (raffinose pentahydrate or trehalose dihydrate) ratio was <1, glycine crystallization was not detected. When the ratio was >or=1, partial glycine crystallization was observed during warming. The presence of amorphous glycine caused the T'g to be substantially lower than that of the solution containing only the carbohydrate. To determine the extent of glycine crystallization, the solutions were annealed for 5 h just above the temperature of glycine crystallization. The T'g observed in the second warming curve was very close to that of the carbohydrate solution alone, indicating almost complete glycine crystallization. These studies enabled the construction of the water-rich sections of the raffinose-glycine-water and trehalose-glycine-water state diagrams. These diagrams consist of a kinetically stable freeze-concentrated solution and a doubly unstable glassy region, which readily crystallizes during cooling or subsequent warming. In addition, there is an intermediate region, where during the experimental timescale, there appears to be hindered glycine nucleation but unhindered crystal growth. To obtain substantially crystalline glycine in the frozen solutions, the glycine to carbohydrate ratios should be >or=1.

The importance of individual protein molecule dynamics in developing and assessing solid state protein preparations.

Processing protein solutions into the solid state is a common approach for generating stable amorphous protein mixtures that are suitable for long-term storage. Great care is typically given to protecting the protein native structure during the various drying steps that render it into the amorphous solid state. However, many studies illustrate that chemical and physical degradations still occur in spite of this amorphous material having good glassy properties and it being stored at temperatures below its glass transition temperature (Tg). Because of these persistent issues and recent biophysical studies that have refined the debate ascribing meaning to the molecular dynamical transition temperature and Tg of protein molecules, we provide an updated discussion on the impact of assessing and managing localized, individual protein molecule nondiffusive motions in the context of proteins being prepared into bulk amorphous mixtures. Our aim is to bridge the pharmaceutical studies addressing bulk amorphous preparations and their glassy behavior, with the biophysical studies historically focused on the nondiffusive internal protein dynamics and a protein's activity, along with their combined efforts in assessing the impact of solvent hydrogen-bonding networks on local stability. We also provide recommendations for future research efforts in solid-state formulation approaches.

Optimization of a Raman microscopy technique to efficiently detect amorphous-amorphous phase separation in freeze-dried protein formulations.

A confocal Raman microscopic technique was optimized to more efficiently detect amorphous-amorphous phase separation in freeze-dried protein formulations. A Renishaw Raman inVia confocal microscope was used to collect 100-200 µm line maps (2 µm step size) of freeze-dried protein-excipient formulations. At each point across the line map, the composition was evaluated from the intensity of the nonoverlapping peaks representative of each component. Collection aperture, scan time, and line map length significantly contributed to the phase-separation analysis, whereas different sample preparation methods did not affect the analysis. Using the optimized parameters (i.e., large aperture 5 s scan time, 200 µm line map), phase separation was successfully detected in binary polymer formulations and was comparable to the previously developed Raman method. However, the previous method required 2.5 h/sample, whereas the optimized method only requires 0.5 h/sample. Phase separation was detected in the following protein-excipient formulations: lysozyme-trehalose (1:1), lysozyme-isomaltose (1:1), ß-lactoglobulin-dextran (1:1), ß-lactoglobulin-dextran (1:3), and ß-lactoglobulin-trehalose (1:1). Phase separation was not detected in lysozyme-sucrose (1:1) and ß-lactoglobulin-sucrose (1:1) formulations. The optimized method successfully detected phase separation in several protein formulations, where phase separation was previously suspected, and promised to be a useful tool for detection of phase separation in amorphous therapeutic formulations.

Water clusters in amorphous pharmaceuticals.

Amorphous materials, although lacking the long-range translational and rotational order of crystalline and liquid crystalline materials, possess certain local (short-range) structure. This paper reviews the distribution of one particular component present in all amorphous pharmaceuticals, that is, water. Based on the current understanding of the structure of water, water molecules can exist in either unclustered form or as aggregates (clusters) of different sizes and geometries. Water clusters are reported in a range of amorphous systems including carbohydrates and their aqueous solutions, synthetic polymers, and proteins. Evidence of water clustering is obtained by various methods that include neutron and X-ray scattering, molecular dynamics simulation, water sorption isotherm, concentration dependence of the calorimetric Tg , dielectric relaxation, and nuclear magnetic resonance. A review of the published data suggests that clustering depends on water concentration, with unclustered water molecules existing at low water contents, whereas clusters form at intermediate water contents. The transition from water clusters to unclustered water molecules can be expected to change water dependence of pharmaceutical properties, such as rates of degradation. We conclude that a mechanistic understanding of the impact of water on the stability of amorphous pharmaceuticals would require systematic studies of water distribution and clustering, while such investigations are lacking.

Impact of sertraline salt form on the oxidative stability in powder blends.

Oxidation of active pharmaceutical ingredients is a common chemical degradation process occurring in solid dosage forms. The aim of this study was to investigate the tendency of various sertraline salts to oxidize in powder blends containing a basic additive. A different extent of conversion of each salt to the free base was observed to occur in the presence of the basic additive, consistent with their respective pHmax values. Sertraline was found to undergo oxidation as the unioinized form, in both solution and powder blends that incorporated an oxidizing agent. In contrast, the ionized form of sertraline remained stable in both cases. Three sertraline salts undergoing a significant extent of conversion from salt to free form in the presence of tribasic sodium phosphate were found to oxidize extensively while sertraline benzoate which had a considerably lower extent of free base formation was more resistant to oxidation. The oxidative degradants were produced through oxidation at the amine functional group of sertraline which is where sertraline is ionized as the salt form. The link between oxidation tendency and the ionization state of sertraline in powder mixtures has thus been demonstrated in this study.

Chemical stability of amorphous materials: specific and general media effects in the role of water in the degradation of freeze-dried zoniporide.

The objective of the present work was to determine whether hydrolysis in a model lyophile was influenced by general media effects with water-changing properties of the medium or via a specific mechanism of water as a reactant. Four formulations of zoniporide and sucrose (1:10) were prepared with variable amounts of sorbitol [0%-25% (w/v) of total solids). These formulations were then equilibrated at 6% and 11% relative humidity using saturated salt solutions. The lyophile cakes were analyzed by differential scanning calorimetery (DSC), (isothermal microcalorimetry (IMC), solid- state nuclear magnetic resonance (ssNMR) spectroscopy, and ultraviolet-visible diffuse reflectance (DFR) spectroscopy. DSC and IMC were used to assess the global molecular mobility. ssNMR relaxation times were measured to access local mobility. The DFR was used to determine the solid-state acidity expressed as the Hammett acidity function. Stability of samples was evaluated at 40°C by monitoring potency and purity by high-performance liquid chromatography (HPLC). Results were interpreted in terms of the various roles of water: media effect, plasticization, polarity, and reactant. The kinetics of hydrolysis was observed to be correlated with either/both specific "chemical" effects, that is, water reactant as well as media effect, specifically global molecular mobility of the matrix. Increase in reaction rate with increase in water content is not linear and is a weaker dependence than in some hydrolytic reactions in organic solvents. A moderate amount of an inert plasticizer, sorbitol, conferred additional stabilization, possibly by restricting the amplitude and frequency of fast motions that are on a small length scale.

Pronounced microheterogeneity in a sorbitol-water mixture observed through variable temperature neutron scattering.

In this study, the structure of concentrated d-sorbitol-water mixtures is studied by wide- and small-angle neutron scattering (WANS and SANS) as a function of temperature. The mixtures are prepared using both deuterated and regular sorbitol and water at a molar fraction of sorbitol of 0.19 (equivalent to 70% by weight of regular sorbitol in water). Retention of an amorphous structure (i.e., absence of crystallinity) is confirmed for this system over the entire temperature range, 100-298 K. The glass transition temperature, Tg, is found from differential scanning calorimetry to be approximately 200 K. WANS data are analyzed using empirical potential structure refinement, to obtain the site-site radial distribution functions (RDFs) and coordination numbers. This analysis reveals the presence of nanoscaled water clusters surrounded by (and interacting with) sorbitol molecules. The water clusters appear more structured compared to bulk water and, especially at the lowest temperatures, resemble the structure of low-density amorphous ice (LDA). Upon cooling to 100 K the peaks in the water RDFs become markedly sharper, with increased coordination number, indicating enhanced local (nanometer-scale) ordering, with changes taking place both above and well below the Tg. On the mesoscopic (submicrometer) scale, although there are no changes between 298 and 213 K, cooling the sample to 100 K results in a significant increase in the SANS signal, which is indicative of pronounced inhomogeneities. This increase in the scattering is partly reversed during heating, although some hysteresis is observed. Furthermore, a power law analysis of the SANS data indicates the existence of domains with well-defined interfaces on the submicrometer length scale, probably as a result of the appearance and growth of microscopic voids in the glassy matrix. Because of the unusual combination of small and wide scattering data used here, the present results provide new physical insight into the structure of aqueous glasses over a broad temperature and length scale, leading to an improved understanding of the mechanisms of temperature- and water-induced (de)stabilization of various systems, including proteins, pharmaceuticals, and biological objects.

Crystalline, liquid crystalline, and isotropic phases of sodium deoxycholate in water.

Sodium deoxycholate (NaDC) is an important example of bile salts, representing systems with complex phase behavior involving both crystalline and mesophase structures. In this study, properties of NaDC-water mixtures were evaluated as a function of composition and temperature via X-ray diffraction with synchrotron (sXRD) and laboratory radiation sources, water sorption, polarized light, hot-stage microscopy, and freezing-point osmometry. Several phases were detected depending on the composition and temperature, including isotropic solution phase, liquid crystalline (LC) phase, crystalline hydrate, and ice. The LC phase was identified as hexagonal structure by sXRD, with up to 14 high-order reflections detected. The crystalline phase was found to be nonstoichiometric hydrate, based on XRD and water sorption data. The phase diagram of NaDC-water system has been refined based on both results of this study and other reports in literature.

Investigation of PEG crystallization in frozen and freeze-dried PEGylated recombinant human growth hormone-sucrose systems: implications on storage stability.

The objectives of the current study were to investigate (i) the phase behavior of a PEGylated recombinant human growth hormone (PEG-rhGH, ∼60 kDa) during freeze-drying and (ii) its storage stability. The phase transitions during freeze-thawing of an aqueous solution containing PEG-rhGH and sucrose were characterized by differential scanning calorimetry. Finally, PEG-rhGH and sucrose formulations containing low, medium, and high polyethylene glycol (PEG) to sucrose ratios were freeze-dried in dual-chamber syringes and stored at 4°C and 25°C. Chemical decomposition (methionine oxidation and deamidation) and irreversible aggregation were characterized by size-exclusion and ion-exchange chromatography, and tryptic mapping. PEG crystallization was facilitated when it was covalently linked with rhGH. When the solutions were frozen, phase separation into PEG-rich and sucrose-rich phases facilitated PEG crystallization and the freeze-dried cake contained crystalline PEG. Annealing caused PEG crystallization and when coupled with higher drying temperatures, the primary drying time decreased by up to 51%. When the freeze-dried cakes were stored at 4°C, while there was no change in the purity of the PEG-rhGH monomer, deamidation was highest in the formulations with the lowest PEG to sucrose ratio. When stored at 25°C, this composition also showed the most pronounced decrease in monomer purity, the highest level of aggregation, and deamidation. Furthermore, an increase in PEG crystallinity during storage was accompanied by a decrease in PEG-rhGH stability. Interestingly, during storage, there was no change in PEG crystallinity in formulations with medium and high PEG to sucrose ratios. Although PEG crystallization during freeze-drying did not cause protein degradation, crystallization during storage might have influenced protein stability.