Recommended Best Practices in Freeze Dryer Equipment Performance Qualification: 2022

Best practices for performing freeze dryer equipment qualification are recommended, focusing on identifying methods to quantify shelf thermal uniformity (also known as "shelf surface uniformity"), equipment capability, and performance metrics of the freeze dryer essential to the pharmaceutical Quality by Design paradigm. Specific guidelines for performing shelf temperature mapping, freeze dryer equipment limit testing (the capability curve), and condenser performance metrics have been provided. Concerning shelf temperature mapping and equipment capability measurements, the importance of paying attention to the test setup and the use of appropriate testing tools are stressed. In all the guidelines provided, much attention has been paid to identifying the balance between obtaining useful process knowledge, logistical challenges associated with testing in the production environment vs that at laboratory scale, and the frequency of the testing necessary to obtain such useful information. Furthermore, merits and demerits of thermal conditions maintained on the cooled surfaces of the freeze dryer condenser have been discussed identifying the specific influence of the condenser surface temperature on the process conditions using experimental data to support the guidelines. Finally, guidelines for systematic leak rate testing criteria for a freeze dryer are presented. These specific procedural recommendations are based on calculations, measurements, and experience to provide useful process and equipment knowledge.

The graphical design space for the primary drying phase of freeze Drying: Factors affecting the dried product layer resistance

An emerging approach to process development of a lyophilized pharmaceutical product is to construct a graphical design space for primary drying as an aid to process optimization. The purpose of this paper is to further challenge the assumption in earlier work that the maximum values of the resistance of dried product layer, R_p, is approximately constant and is independent of process conditions within the "acceptable" region of the design space. Three model formulations containing bovine serum albumin as the model protein were chosen to represent: (a) an amorphous system, (b) a crystalline system, and (c) a mixed system where both an amorphous and a crystalline component were present. Low temperature differential scanning calorimetry (DSC) and freeze dry microscopy (FDM) experiments were conducted to estimate critical product temperature. A conservative lyophilization cycle was conducted for each formulation to collect mass flow data and individual design spaces were then established. A series of lyophilization cycles were then conducted using process conditions that resided within the individual design space and the resultant product temperature and resistance of dried product layer (R_p) values were compared between the individual cycles within each formulation. The data indicated that the R_p was component dependent with the mannitol formulation exhibiting higher R_p values than the sucrose formulation. Interestingly, when mannitol was retained amorphous, the formulation exhibited a lower R_p, similar to that of the sucrose formulation. The mixed formulation exhibited intermediate R_p values. Crystallization of mannitol is hypothesized to facilitate a decrease in the size of the ice porous structure by making the water vapor flow path tortuous, thereby increasing the R_p of mannitol formulations. Within the "acceptable" zone of the individual design space, R_p was dependent on the process condition with more aggressive shelf temperature cycles resulting in lower R_p. Specific Surface Area measurements of freeze-dried solids demonstrated that more aggressive conditions resulted in smaller surface area. Freeze-dried solids of crystalline formulations consistently exhibited higher specific surface area than the amorphous formulations.

Stability of antibody drug conjugate formulations evaluated using solid-state hydrogen-deuterium exchange mass spectrometry

Antibody drug conjugates (ADCs) have been at the forefront in cancer therapy due to their target specificity. All the FDA approved ADCs are developed in lyophilized form to minimize instability associated with the linker that connects the cytotoxic drug and the antibody during shipping and storage. We present here solid-state hydrogen-deuterium exchange with mass spectrometric analysis (ssHDX-MS) as a tool to analyze protein structure and matrix interactions for formulations of an ADC with and without commonly used excipients. We compared results of the ssHDX-MS with accelerated stability results using size-exclusion chromatography and determined that the former technique was able to successfully identify the destabilizing effects of mannitol and polysorbate 80. In comparison, Fourier-transform infrared spectroscopy results were inconclusive. The agreement between ssHDX-MS and stressed stability studies supports the potential of ssHDX-MS as a method of predicting relative stability of different formulations.

Equipment Capability Measurement of Laboratory Freeze-Dryers: a Comparison of Two Methods

The objective of this investigation was to evaluate two methods for measuring the maximum sublimation rate that a freeze-dryer will support-the minimum controllable pressure method and the choke point method. Both methods gave equivalent results, but the minimum controllable pressure method is preferred, since it is easier, faster, and less subjective. The ratio of chamber pressure to condenser pressure corresponding to the onset of choked flow was considerably higher in this investigation (up to about 20:1) than in previously published reports. This ratio was not affected by the location of the pressure gauge on the condenser; that is, on the foreline of the vacuum pump versus on the body of the condenser itself. The total water loss due to sublimation as measured by tunable diode laser absorption spectroscopy was consistently within 5% of gravimetrically determined weight loss, regardless of whether the measurement took place during choked versus non-choked process conditions.

Freeze-Dryer Equipment Capability Limit: Comparison of Computational Modeling With Experiments at Laboratory Scale

The equipment capability curve is one of the bounding elements of the freeze-drying design space, and understanding it is critical to process design, transfer, and scale-up. The second bounding element of the design space is the product temperature limit beyond which the product collapses. The high cost associated with freeze-drying any product renders it crucial to operate using the most efficient cycle within the limits of the equipment and the product. In this work, we present a computational model to generate the equipment capability curve for 2 laboratory scale freeze-dryers and compare the results to experimentally generated equipment capability curves. The average deviations of the modeling results from the experiments for the 2 lyophilizers modeled are -4.8% and -7.2%. In addition, we investigate the effect of various numerical and geometric parameters on the simulated equipment capability. Among the numerical parameters, the chamber wall thermal boundary conditions exert the largest influence with a maximum value of 12.3%. Among the geometric parameters, the inclusion of the isolation valve reduces the equipment capability by 23.7%. Larger isolation valves, required for controlled nucleation technology, choke the flow in the duct at lower sublimation rates, thereby lowering the equipment capability limit.

Experimental Aspects of Measuring the Vial Heat Transfer Coefficient in Pharmaceutical Freeze-Drying

One of the current methods for cycle optimization in primary drying to is develop a graphical design space based on quality by design (QbD). In order to construct the design space, the vial heat transfer coefficient (K_v) is needed. This paper investigated experimental factors that can affect the K_v result, examined the relationship between the batch average K_v and K_v values for individual vials, and recommended best practices for measuring K_v. Factors investigated included the technique for measuring ice temperature, shelf temperature, the use of a radiation shield on the door of the freeze-dry chamber, and shelf spacing. All experiments reported here used a chamber pressure of 100 mTorr. The most important factor was the technique for ice temperature measurement, where it is important to assure that any restrictions to vapor flow at the top of the vial are the same between monitored and non-monitored vials. Another factor that was found to play a role was the shelf temperature whereby the lower the shelf temperature, the larger the "edge effect," and the larger the average K_v. Factors that were found to not have a significant effect were the use of a radiation shield inside the chamber door and the shelf spacing. Being aware of these factors and knowing best practices when determining the vial heat coefficient will lead to more accurate design spaces and better cycle optimization.

Process and Formulation Effects on Protein Structure in Lyophilized Solids Using Mass Spectrometric Methods

Myoglobin (Mb) was lyophilized in the absence (Mb-A) and presence (Mb-B) of sucrose in a pilot-scale lyophilizer with or without controlled ice nucleation. Cake morphology was characterized using scanning electron microscopy, and changes in protein structure were monitored using solid-state Fourier-transform infrared spectroscopy, solid-state hydrogen-deuterium exchange-mass spectrometry, and solid-state photolytic labeling-mass spectrometry (ssPL-MS). The results showed greater variability in nucleation temperature and irregular cake structure for formulations lyophilized without controlled nucleation. Controlled nucleation resulted in nucleation at ∼(-5°C) and uniform cake structure. Formulations containing sucrose showed better retention of protein structure by all measures than formulations without sucrose. Samples lyophilized with and without controlled nucleation were similar by most measures of protein structure. However, ssPL-MS showed the greatest photoleucine incorporation and more labeled regions for Mb-B lyophilized with controlled nucleation. The data support the use of solid-state hydrogen-deuterium exchange-mass spectrometry and ssPL-MS to study formulation and process-induced conformational changes in lyophilized proteins.

Influence of ethanol on physical state of freeze-dried mannitol

PURPOSE

The purpose of this study is to characterize freeze-dried mannitol prepared from an ethanol-containing solution as a function of the ethanol ratio, mannitol concentration, and annealing in the freeze-drying cycle.

METHODS

The characteristics of the freeze-dried mannitol were evaluated by X-ray diffractometry (XRD) and differential scanning calorimetry (DSC). The reconstitution time was measured for the freeze-dried solids as well as the residual moisture and ethanol by Karl-Fischer titration and gas chromatography, respectively.

RESULTS

The XRD pattern of 5% (w/v) mannitol freeze-dried from aqueous solution with no annealing cycle showed all the five characteristic peaks at 13.6 degrees and 17.2 degrees 2theta for the alpha polymorph, at 14.6 degrees and 23.4 degrees 2theta for the beta polymorph and at 9.7 degrees 2theta for the delta polymorph. The addition of ethanol to the initial solutions resulted in only a peak at 9.7 degrees 2theta, indicating the presence of only the delta polymorph, regardless of the ethanol ratio in the initial solutions used [10, 20, 30, and 40% (v/v)]. However, annealing during freeze-drying influenced the XRD pattern; in particular, for the solid prepared from the 10% ethanol solution. Annealing of the 10% ethanol solution promoted the formation of the alpha polymorph and produced a different peak that might be attributable to another polymorph. In DSC thermograms, an endotherm and a subsequent exotherm were found in the temperature range of 150 degrees C to 160 degrees C, which corresponded to the transition of the delta form to alpha or beta forms. The magnitude of this transition was smaller as the ethanol ratio increased for the solids from ethanol-containing solutions with an annealing cycle. In other words, annealing of the ethanol-containing solutions promoted delta polymorph formation in the lyophiles. In addition, the mannitol concentration affected the polymorphism in freeze-dried solids prepared from aqueous and 10% ethanol solutions. Addition of ethanol in the initial solution, in particular, at a lower ethanol level (10% v/v), and a higher concentration of mannitol could also promote the generation of lumps in freeze-dried solids during reconstitution, and result in longer reconstitution time. The residual moisture levels were less than 0.5%, and residual ethanol levels were less than 0.1%, irrespective of the formulation used.

CONCLUSIONS

The physical state and reconstitution time of the freeze-dried mannitol appears to be a complex function of the ethanol and mannitol concentrations in the initial solution before freeze-drying and of annealing during the freeze-drying process.

Subambient behavior of mannitol in ethanol-water co-solvent system

PURPOSE

The purpose of this study is to characterize the freezing behavior of mannitol in ethanol-water co-solvent systems in comparison with the corresponding aqueous solution.

METHODS

Subambient differential scanning calorimetry (DSC) and microscopy techniques were used to investigate the freezing behavior of mannitol in aqueous solutions and in ethanol-water co-solvent systems.

RESULTS

The DSC thermogram of the frozen aqueous solution, which was warmed after cooling at 5.0 degrees C/min, consisted of a glass transition, an endothermic transition, and a crystallization exotherm from mannitol, respectively. The thermograms of ethanol-containing solutions were different in view of including some thermal events attributable to ethanol hydrates. The glass transition of amorphous mannitol was also observed in the thermograms, but became unclear with increasing ethanol in the co-solvent system. The microscopy experiments enabled understanding of the subambient behavior of mannitol. Ethanol was largely removed by vacuum drying rather than freeze-drying. In addition, such manipulations as annealing during the freezing process and slower cooling (0.5 degrees C/min) enhanced the crystallization of mannitol in the frozen system.

CONCLUSIONS

In the presence of ethanol, crystallization of mannitol was inhibited under subambient conditions. Annealing or slower cooling promoted the crystallization of mannitol during the freezing process.

Identification of Critical Process Variables Affecting Particle Size Following Precipitation Using a Supercritical Fluid

The critical processing parameters affecting average particle size, particle size distribution, yield, and level of residual carrier solvent using the supercritical anti-solvent method (SAS) were identified. Carbon dioxide was used as the supercritical fluid. Methylprednisolone acetate was used as the model solute in tetrahydrofuran. Parameters examined included pressure of the supercritical fluid, agitation rate, feed solution flow rate, impeller diameter, and nozzle design. Pressure was identified as the most important process parameter affecting average particle size, either through the effect of pressure on dispersion of the feed solution into the precipitation vessel or through the effect of pressure on solubility of drug in the CO2/organic solvent mixture. Agitation rate, impeller diameter, feed solution flow rate, and nozzle design had significant effects on particle size, which suggests that dispersion of the feed solution is important. Crimped HPLC tubing was the most effective method of introducing feed solution into the precipitation vessel, largely because it resulted in the least amount of clogging during the precipitation. Yields of 82% or greater were consistently produced and were not affected by the processing variables. Similarly, the level of residual solvent was independent of the processing variables and was present at 0.0002% wt/wt THF or less.

Identification of Physical-Chemical Variables Affecting Particle Size Following Precipitation Using a Supercritical Fluid

The physical-chemical processing variables affecting particle size following precipitation using the supercritical antisolvent (SAS) method were investigated by varying both the composition of the feed solvent and the structure of the solute, using a series of steroids. The key factor influencing particle size in these studies appears to be the solubility of the drug in the organic solvent/supercritical fluid mixture, where relatively high solubility causes a lower degree of supersaturation in the precipitation vessel, resulting in a relatively large particle size. Higher operating pressures result in larger particle sizes, probably through the effect of pressure on solubility. Physical properties of the carrier solvent, such as vapor pressure and dielectric constant, were not effective predictors of relative particle size of the precipitated powder, nor was solubility of the model drug in the carrier solvent. In limited studies of the physical state of the precipitated solid, higher apparent crystallinity was observed for powders with larger particle size. A precipitate of a different crystal form was observed when starting with hydrocortisone hemisuccinate monohydrate and may represent the loss of water of hydration. An amorphous solid was precipitated when starting with yttrium acetate dihydrate. Broad guidelines for effective particle size reduction using this technique are presented.

Top ten hot topics in parenteral science and technology

Ten current "hot topics" in parenteral science and technology are reviewed to update the reader on current advances and challenges with each topic. Topics selected are formulation advances, packaging advances, extractables and leachables, analytical method advances for biopharmaceuticals, protein pharmaceutics, quality by design, manufacturing and equipment advances, aseptic processing advances, rapid microbial methods, and visual inspection of parenteral products.

Evaluation of manometric temperature measurement, a process analytical technology tool for freeze-drying: part II measurement of dry-layer resistance

The purpose of this work was to study the factors that may cause systematic errors in the manometric temperature measurement (MTM) procedure used to determine product dry-layer resistance to vapor flow. Product temperature and dry-layer resistance were obtained using MTM software installed on a laboratory freeze-dryer. The MTM resistance values were compared with the resistance values obtained using the "vial method." The product dry-layer resistances obtained by MTM, assuming fixed temperature difference (DeltaT; 2 degrees C), were lower than the actual values, especially when the product temperatures and sublimation rates were low, but with DeltaT determined from the pressure rise data, more accurate results were obtained. MTM resistance values were generally lower than the values obtained with the vial method, particularly whenever freeze-drying was conducted under conditions that produced large variations in product temperature (ie, low shelf temperature, low chamber pressure, and without thermal shields). In an experiment designed to magnify temperature heterogeneity, MTM resistance values were much lower than the simple average of the product resistances. However, in experiments where product temperatures were homogenous, good agreement between MTM and "vial-method" resistances was obtained. The reason for the low MTM resistance problem is the fast vapor pressure rise from a few "warm" edge vials or vials with low resistance. With proper use of thermal shields, and the evaluation of DeltaT from the data, MTM resistance data are accurate. Thus, the MTM method for determining dry-layer resistance is a useful tool for freeze-drying process analytical technology.

Evaluation of manometric temperature measurement (MTM), a process analytical technology tool in freeze drying, part III: heat and mass transfer measurement

This article evaluates the procedures for determining the vial heat transfer coefficient and the extent of primary drying through manometric temperature measurement (MTM). The vial heat transfer coefficients (Kv) were calculated from the MTM-determined temperature and resistance and compared with Kv values determined by a gravimetric method. The differences between the MTM vial heat transfer coefficients and the gravimetric values are large at low shelf temperature but smaller when higher shelf temperatures were used. The differences also became smaller at higher chamber pressure and smaller when higher resistance materials were being freeze-dried. In all cases, using thermal shields greatly improved the accuracy of the MTM Kv measurement. With use of thermal shields, the thickness of the frozen layer calculated from MTM is in good agreement with values obtained gravimetrically. The heat transfer coefficient "error" is largely a direct result of the error in the dry layer resistance (ie, MTM-determined resistance is too low). This problem can be minimized if thermal shields are used for freeze-drying. With suitable use of thermal shields, accurate Kv values are obtained by MTM; thus allowing accurate calculations of heat and mass flow rates. The extent of primary drying can be monitored by real-time calculation of the amount of remaining ice using MTM data, thus providing a process analytical tool that greatly improves the freeze-drying process design and control.

Infrared microscopy for in situ measurement of protein secondary structure during freezing and freeze-drying

A commercially available freeze-dry microscopy stage interfaced with an IR microscope is described as a method of in situ measurement of protein secondary structure in the liquid, frozen and freeze-dried states. Studies using solutions of model proteins demonstrated that spectra collected using the IR microscope have resolution and sensitivity that is comparable to techniques using a conventional infrared spectrometer. Additionally, spectra collected in triplicate on the microscope in the solution, frozen, and freeze-dried states and after reconstitution were shown to be reproducible. The limiting factor when collecting spectra on the infrared microscope appears to be the higher level of water vapor inherently present within the optical path of the microscope used in this study. Results demonstrate that the native secondary structure is perturbed in both the frozen and freeze-dried states, and bands characteristic of structural changes associated with freezing and drying stresses were observed in the Amide I region. Freeze-drying studies conducted in the presence of mannitol and sucrose demonstrated that perturbation to the native state secondary structure after freeze-drying was considerably reduced in the presence of these excipients.

Nuclear Magnetic Resonance Imaging of Freeze-Drying

The objective of this study was to determine the feasibility of using magnetic resonance imaging (MRI) to observe the shape and position of the ice sublimation front during primary drying for subsequent comparison of these images with those predicted by mathematical model of heat and mass transfer during primary drying. One-dimensional (1D) profile studies on both ice and 2% HSA during in situ freeze-drying were performed with a single point imaging (SPI) sequence, and demonstrated that the SPI technique can be used in capturing the ice signal during freeze-drying. In order to perform two-dimensional imaging, it was found that the ice must be "doped" in order to increase the apparent transverse relaxation time (T2*) and decrease the longitudinal relaxation time (T1) of ice. HBr at a level of 0.1 mM was shown to be an effective dopant. As a result, 2D images of ice and 5% HSA doped with 0.1 mM HBr were successfully collected, and the shape and position of the ice sublimation front during in situ freeze-drying were observed.

Evaluation of manometric temperature measurement, a process analytical technology tool for freeze-drying: part I, product temperature measurement

This study examines the factors that may cause systematic errors in the manometric temperature measurement (MTM) procedure used to evaluate product temperature during primary drying. MTM was conducted during primary drying using different vial loads, and the MTM product temperatures were compared with temperatures directly measured by thermocouples. To clarify the impact of freeze-drying load on MTM product temperature, simulation of the MTM vapor pressure rise was performed, and the results were compared with the experimental results. The effect of product temperature heterogeneity in MTM product temperature determination was investigated by comparing the MTM product temperatures with directly measured thermocouple product temperatures in systems differing in temperature heterogeneity. Both the simulated and experimental results showed that at least 50 vials (5 mL) were needed to give sufficiently rapid pressure rise during the MTM data collection period (25 seconds) in the freeze dryer, to allow accurate determination of the product temperature. The product temperature is location dependent, with higher temperature for vials on the edge of the array and lower temperature for the vials in the center of the array. The product temperature heterogeneity is also dependent upon the freeze-drying conditions. In product temperature heterogeneous systems, MTM measures a temperature close to the coldest product temperature, even if only a small fraction of the samples have the coldest product temperature. The MTM method is valid even at very low product temperature (-45 degrees C).

Freeze-drying of proteins from a sucrose-glycine excipient system: effect of formulation composition on the initial recovery of protein activity

The purpose of this study was to investigate the effect of sucrose-glycine excipient systems on the stability of selected model proteins during lyophilization. Recovery of protein activity after freeze-drying was examined for the model proteins lactate dehydrogenase and glucose 6-phosphate dehydrogenase in a sucrose-glycine-based excipient system in which the formulation composition was systematically varied. In a sucrose-only excipient system, activity recovery of both model proteins is about 80% and is independent of sucrose concentration over a range from 1 to 40 mg/mL. When both sucrose and glycine are used and the ratio of the 2 excipients is varied, however, activity recovery decreases in a pattern that is consistent with the inhibition of activity recovery by glycine crystals, despite the presence of an adequate amount of sucrose to afford protection. Annealing of sucrose-glycine formulations causes a small but significant decrease in activity recovery relative to unannealed controls, whereas no annealing effect is observed with sucrose-only formulations. Addition of 0.01% polysorbate 80 to the formulation resulted in complete recovery of activity, irrespective of the sucrose-glycine ratio or annealing. Addition of the same concentration of polysorbate 80 to the reconstitution medium caused an increase in activity recovery for each formulation, but the overall pattern remained unchanged. The data are consistent with an interfacial model for lyophilization-associated loss of protein activity involving denaturation at a solid/freeze-concentrate interface.

Freeze-Drying Process Design by Manometric Temperature Measurement: Design of a Smart Freeze-Dryer

PURPOSE

To develop a procedure based on manometric temperature measurement (MTM) and an expert system for good practices in freeze drying that will allow development of an optimized freeze-drying process during a single laboratory freeze-drying experiment.

METHODS

Freeze drying was performed with a FTS Dura-Stop/Dura-Top freeze dryer with the manometric temperature measurement software installed. Five percent solutions of glycine, sucrose, or mannitol with 2 ml to 4 ml fill in 5 ml vials were used, with all vials loaded on one shelf. Details of freezing, optimization of chamber pressure, target product temperature, and some aspects of secondary drying are determined by the expert system algorithms. MTM measurements were used to select the optimum shelf temperature, to determine drying end points, and to evaluate residual moisture content in real-time. MTM measurements were made at 1 hour or half-hour intervals during primary drying and secondary drying, with a data collection frequency of 4 points per second. The improved MTM equations were fit to pressure-time data generated by the MTM procedure using Microcal Origin software to obtain product temperature and dry layer resistance. Using heat and mass transfer theory, the MTM results were used to evaluate mass and heat transfer rates and to estimate the shelf temperature required to maintain the target product temperature.

RESULTS

MTM product dry layer resistance is accurate until about two-thirds of total primary drying time is over, and the MTM product temperature is normally accurate almost to the end of primary drying provided that effective thermal shielding is used in the freeze-drying process. The primary drying times can be accurately estimated from mass transfer rates calculated very early in the run, and we find the target product temperature can be achieved and maintained with only a few adjustments of shelf temperature. The freeze-dryer overload conditions can be estimated by calculation of heat/mass flow at the target product temperature. It was found that the MTM results serve as an excellent indicator of the end point of primary drying. Further, we find that the rate of water desorption during secondary drying may be accurately measured by a variation of the basic MTM procedure. Thus, both the end point of secondary drying and real-time residual moisture may be obtained during secondary drying.

CONCLUSIONS

Manometric temperature measurement and the expert system for good practices in freeze drying does allow development of an optimized freeze-drying process during a single laboratory freeze-drying experiment.

Kinetics of glycine crystallization during freezing of sucrose/glycine excipient systems

Isothermal calorimetry and low-field nuclear magnetic resonance were used to measure crystallization of glycine during annealing of glycine/sucrose mixtures, a commonly-used excipient system for freeze-dried proteins. Kinetics of crystallization of glycine were consistent between the two methods, although the NMR method was significantly more sensitive. By the calorimetric method used here, sensitivity was lost when the total solute concentration was below about 20% (w/v) and the relative glycine concentration below about 35% of the total solids. By the NMR method, total solute concentrations as low as 5% (w/v) could be studied. When the relative concentration of glycine is below about 30% of total solids, the time course of crystallization becomes excessively long for practical freeze-drying applications. A good fit of the crystallization data was obtained with the Johnson-Mehl-Avrami (JMA) equation. The Avrami exponent of 2.5 is consistent with diffusion-limited spherulitic growth of glycine.

Effect of Collapse on the Stability of Freeze-Dried Recombinant Factor VIII and α-amylase

Recombinant Factor VIII (rFVIII) and alpha-amylase were used as model proteins to examine the effect of freeze-drying process conditions on the long-term stability of these proteins as freeze-dried solids. The same sucrose/glycine formulation was used for all treatments. Three freeze-drying protocols were used-an "aggressive" and a "conservative" cycle that both produced pharmaceutically acceptable product, and a protocol that produced a collapsed matrix. For rFVIII, there was no difference in the biological activity versus the time profile for product freeze-dried under the three different conditions when stored at 5 or 25 degrees C. At 40 degrees C, however, the stability of the collapsed product appeared to be better than that of product freeze-dried with no collapse. Also, the level of residual moisture in the collapsed product was higher than that of the product with no collapse. For alpha-amylase, there was no significant difference in the stability profile at any of the temperatures over the time course of the study. The results support the conclusion that collapse is not necessarily detrimental to the long-term stability of freeze-dried proteins.

Fundamentals of freeze-drying

Given the increasing importance of reducing development time for new pharmaceutical products, formulation and process development scientists must continually look for ways to "work smarter, not harder." Within the product development arena, this means reducing the amount of trial and error empiricism in arriving at a formulation and identification of processing conditions which will result in a quality final dosage form. Characterization of the freezing behavior of the intended formulation is necessary for developing processing conditions which will result in the shortest drying time while maintaining all critical quality attributes of the freeze-dried product. Analysis of frozen systems was discussed in detail, particularly with respect to the glass transition as the physical event underlying collapse during freeze-drying, eutectic mixture formation, and crystallization events upon warming of frozen systems. Experiments to determine how freezing and freeze-drying behavior is affected by changes in the composition of the formulation are often useful in establishing the "robustness" of a formulation. It is not uncommon for seemingly subtle changes in composition of the formulation, such as a change in formulation pH, buffer salt, drug concentration, or an additional excipient, to result in striking differences in freezing and freeze-drying behavior. With regard to selecting a formulation, it is wise to keep the formulation as simple as possible. If a buffer is needed, a minimum concentration should be used. The same principle applies to added salts: If used at all, the concentration should be kept to a minimum. For many proteins a combination of an amorphous excipient, such as a disaccharide, and a crystallizing excipient, such as glycine, will result in a suitable combination of chemical stability and physical stability of the freeze-dried solid. Concepts of heat and mass transfer are valuable in rational design of processing conditions. Heat transfer by conduction--the dominant mechanism of heat transfer in freeze-drying--is inefficient at the pressures used in freeze-drying. Steps should be taken to improve the thermal contact between the product and the shelf of the freeze dryer, such as eliminating metal trays from the drying process. Quantitation of the heat transfer coefficient for the geometry used is a useful way of assessing the impact of changes in the system such as elimination of product trays and changes in the vial. Because heat transfer by conduction through the vapor increases with increasing pressure, the commonly held point of view that "the lower the pressure, the better" is not true with respect to process efficiency. The optimum pressure for a given product is a function of the temperature at which freeze-drying is carried out, and lower pressures are needed at low product temperatures. The controlling resistance to mass transfer is almost always the resistance of the partially dried solids above the submination interface. This resistance can be minimized by avoiding fill volumes of more than about half the volume of the container. The development scientist should also recognize that very high concentrations of solute may not be appropriate for optimum freeze-drying, particularly if the resistance of the dried product layer increases sharply with concentration. Although the last 10 years has seen the publication of a significant body of literature of great value in allowing development scientists and engineers to "work smarter," there is still much work needed in both the science and the technology of freeze-drying. Scientific development is needed for improving analytical methodology for characterization of frozen systems and freeze-dried solids. A better understanding of the relationship between molecular mobility and reactivity is needed to allow accurate prediction of product stability at the intended storage temperature based on accelerated stability at higher temperatures. This requires that the temperature dependence of glass transition-associated mobility, particularly at temperatures below the glass transition, be studied in greater depth. The relevance of the concept of strong and fragile glasses to frozen systems and freeze-dried solids has only begun to be explored. The list of pharmaceutically acceptable protective solutes is very short, and more imagination--and work--is needed in order to develop pharmaceutically acceptable alternative stabilizers. There is a need for technology development in process monitoring, particularly in developing a way to measure the status of the product during freezing and freeze-drying without placing temperature measurement probes in individual vials of product. The current practice of placing thermocouples in vials is uncertain with respect to reliability of the data, inconsistent with elimination of personnel in close proximity to open vials of product in an aseptic environment, and incompatible with technology for automatic material handling in freeze-drying. In addition, a method for controlling the degree of supercooling during freezing would allow better control of freezing rate and would, in many cases, result in more consistent product quality.

Freeze-drying of tert-butanol/water cosolvent systems: a case report on formation of a friable freeze-dried powder of tobramycin sulfate

A case study is presented in which a tert-butanol (TBA)/water cosolvent system was found to be a useful means of producing freeze-dried tobramycin sulfate that readily forms a loose powder upon agitation in a specialized application in which a critical quality attribute is the ability to pour the sterile powder from the vial. Both formulation and processing variables are important in achieving acceptable physical properties of the cake as well as minimizing residual TBA levels. Liquid/liquid phase separation was observed above critical concentrations of both drug and TBA, resulting in a two-layered lyophilized cake with unacceptable appearance, physical properties, and residual TBA levels. However, the choice of tobramycin sulfate and TBA concentrations in the single-phase region of the phase diagram resulted in a lyophilized solid that can readily be poured from vials. Crystallization of TBA before drying is critical to achieving adequately low residual TBA levels, and this is reflected in the effect of thermal history of freezing on residual TBA levels, where rapid freezing results in incomplete crystallization of TBA and relatively high levels of residual solvent. Annealing at a temperature above T'(g) of the system after an initial freezing step significantly reduces the level of residual TBA. Secondary drying, even at increased temperature and for extended times, is not an effective method of reducing residual TBA levels.

Characterization of frozen solutions of glycine

The broad objective of this research was to better understand the physical chemistry of freeze drying of the system glycine/water, with emphasis on the role of polymorphism of glycine on freezing and freeze-drying behavior. Frozen solutions of glycine were characterized by differential scanning calorimetry (DSC) and by freeze-dry microscopy. Cooling rates ranged from 0.1 degrees C/min to quench-cooling by immersing samples in liquid nitrogen. During slow cooling, only a beta-glycine/ice eutectic mixture is formed, melting at -3.60 degrees C. For quench-frozen solutions, the low-temperature thermal behavior is more complex. A complex glass transition region is observed on the DSC thermogram, with midpoint temperatures at about -73 degrees C and -60 degrees C, as well as two separate crystallization exotherms. Use of very low heating rates in the DSC experiment allows resolution of four separate endotherms in the temperature range just below the melting of ice. The experimental data support the conclusion that these endotherms arise from melting of the beta-glycine/ice eutectic mixture at -3.6 degrees C, dissolution of crystals of alpha-glycine at -2.85 degrees C, and melting of the gamma-glycine/ice eutectic mixture at -2.70 degrees C. One of the endotherms could not be characterized because of inadequate resolution from the beta-glycine/ice eutectic melting endotherm. Freeze-dried solids were characterized by X-ray powder diffraction after annealing under conditions established by the DSC and freeze-dry microscopy experiments. Annealing at controlled temperatures in the melting region prior to recooling the system was useful not only in interpreting the complex DSC thermogram, but also in controlling the glycine polymorph resulting from freeze drying.

Degree of antigen adsorption in the vaccine or interstitial fluid and its effect on the antibody response in rabbits

The effect of the degree of adsorption of lysozyme by aluminium hydroxide adjuvant on the immune response in rabbits was studied. The surface charge of the adjuvant was modified by pretreatment with phosphate anion to produce five vaccines having degrees of adsorption ranging from 3 to 90%. The degree of adsorption of vaccines exhibiting 3, 35 or 85% adsorption changed to 40% within 1 h after each vaccine was mixed with sheep interstitial fluid to simulate subcutaneous administration. The mean anti-lysozyme antibody titers produced by the vaccines were the same and were four times greater than that produced by a lysozyme solution. Thus, the degree of adsorption of lysozyme in sheep interstitial fluid rather than the degree of adsorption in the vaccine correlated with the immune response.

Effects of buffer composition and processing conditions on aggregation of bovine IgG during freeze-drying

The objective of this study was to identify critical formulation and processing variables affecting aggregation of bovine IgG during freeze-drying when no lyoprotective solute is used. Parameters examined were phosphate buffer concentration and counterion (Na versus K phosphate), added salts, cooling rate, IgG concentration, residual moisture level, and presence of a surfactant. No soluble aggregates were detected in any formulation after either freezing/thawing or freeze-drying. No insoluble aggregates were detected in any formulation after freezing, but insoluble aggregate levels were always detectable after freeze-drying. The data are consistent with a mechanism of aggregate formation involving denaturation of IgG at the ice/freeze-concentrate interface which is reversible upon freeze-thawing, but becomes irreversible after freeze-drying and reconstitution. Rapid cooling (by quenching in liquid nitrogen) results in more and larger aggregates than slow cooling on the shelf of the freeze-dryer. This observation is consistent with surface area measurements and environmental electron microscopic data showing a higher surface area of freeze-dried solids after fast cooling. Annealing of rapidly cooled solutions results in significantly less aggregation in reconstituted freeze-dried solids than in nonannealed controls, with a corresponding decrease in specific surface area of the freeze-dried, annealed system. Increasing the concentration of IgG significantly improves the stability of IgG against freeze-drying-induced aggregation, which may be explained by a smaller percentage of the protein residing at the ice/freeze-concentrate interface as IgG concentration is increased. A sodium phosphate buffer system consistently results in more turbid reconstituted solids than a potassium phosphate buffer system at the same concentration, but this effect is not attributable to a pH shift during freezing. Added salts such as NaCl or KCl contribute markedly to insoluble aggregate formation. Both sodium and potassium chloride contribute more to turbidity of the reconstituted solid than either sodium or potassium phosphate buffers at similar ionic strength, with sodium chloride resulting in a substantially higher level of aggregates than potassium chloride. At a given cooling rate, the specific surface area of dried solids is approximately a factor of 2 higher for the formulation containing sodium chloride than the formulation containing potassium chloride. Turbidity is also influenced by the extent of secondary drying, which underscores the importance of minimizing secondary drying of this system. Including a surfactant such as polysorbate 80, either in the formulation or in the water used for reconstitution, decreased, but did not eliminate, insoluble aggregates. There was no correlation between pharmaceutically acceptability of the freeze-dried cake and insoluble aggregate levels in the reconstituted product.

Frequency-domain photon migration measurements for quantitative assessment of powder absorbance: A novel sensor of blend homogeneity

The measurement and analysis of frequency-domain photon migration (FDPM) measurements of powder absorbance in pharmaceutical powders is described in the context of other optical techniques. FDPM consists of launching intensity-modulated light into a powder and detecting the phase delay and amplitude modulation of the re-emitted light as a function of the modulation frequency. From analysis of the data using the diffusion approximation to the radiative transport equation, the absorption coefficient can be obtained. Absorption coefficient measurements of riboflavin in lactose mixtures are presented at concentrations of 0.1 to 1% (w/w) at near-infrared wavelengths where solution absorption cross sections are difficult to accurately measure using traditional transmission measurements in nonscattering solutions. FDPM measurements in powders enabled determinations of absorption coefficients that increase linearly with concentration (w/w) according to Beer-Lambert relationship. The extension of FDPM for monitoring absorbance of low-dose and ultralow-dose powder blending operations is presented.

Effects of freeze-dry processing conditions on the crystallization of pentamidine isethionate

The results of this study show that pentamidine isethionate (PI) can exist in at least four crystalline forms-three anhydrates designated as forms A, B, and C, and a trihydrate. Form C is the high-temperature modification, produced by heating forms A, B, and the trihydrate above 130 degrees C and cannot be produced under actual lyophilization conditions. The crystal forms of PI present after freeze-drying depend on the initial solution concentration and the thermal history of freezing. At low concentrations of PI (4% and less), form A is observed regardless of freezing method. At a higher concentration (10%), the crystal forms observed are a function of the freezing method. Three freezing methods were used to effect different cooling rates: (1) cooling on the shelf to 2 degrees C and holding for 3 h prior to decreasing the temperature to -45 degrees C, (2) directly cooling on the shelf from room temperature to -45 degrees C, and (3) dipping the vials in liquid nitrogen. The results show that form A, form B, or a mixture of both forms are present in the freeze-dried solid depending upon whether the trihydrate crystallizes during freezing or not. Since form B can only be produced by dehydration of the trihydrate at low temperature, the presence of this form in the freeze-dried powders depends on the nucleation and growth of the trihydrate during freezing. Photostability studies have demonstrated marked differences between freeze-dried solids frozen under different conditions. The results underscore the importance of recognizing that seemingly subtle differences in processing conditions can have a significant impact on critical quality attributes of freeze-dried products.

The physical state of mannitol after freeze-drying: effects of mannitol concentration, freezing rate, and a noncrystallizing cosolute

The objectives of this study were to (1) measure the effects of freezing rate and mannitol concentration on the physical state of freeze-dried mannitol when mannitol is present as a single component, (2) determine the relative concentration threshold above which crystalline mannitol can be observed by X-ray powder diffraction in the freeze-dried solid when a variety of noncrystallizing solutes are included in the formulation, and (3) measure the glass transition temperature of amorphous mannitol and to determine the degree to which the glass transition temperature of freeze-dried solids consisting of mannitol and a disaccharide is predicted by the Gordon-Taylor equation. Both freezing rate and mannitol concentration influence the crystal form of mannitol in the freeze-dried solid when mannitol is present as a single component. Slow freezing of 10% (w/v) mannitol produces a mixture of the alpha and beta polymorphs, whereas fast freezing of the same solution produces the delta form. Fast freezing of 5% (w/v) mannitol results primarily in the beta form. The threshold concentration above which crystalline mannitol is detected in the freeze-dried solid by X-ray diffraction is consistently about 30% (w/w) when a second, noncrystallizing solute is present, regardless of the nature of the second component. The glass transition temperature of amorphous mannitol measured from the quench-cooled melt is approximately 13 degreesC. Accordingly, mannitol is an effective plasticizer of freeze-dried solids when the mannitol remains amorphous. Glass transition temperatures of mixtures of mannitol and the disaccharides sucrose, maltose, trehalose, and lactose are well predicted by the Gordon-Taylor equation with values of k in the range of 3 to 4.

Effect of process conditions on recovery of protein activity after freezing and freeze-drying

The objective of this research was to gain a better understanding of the degree to which recovery of activity of model proteins after freeze-drying can be maximized by manipulation of freeze-dry process conditions in the absence of protective solutes. Catalase, beta-galactosidase and lactate dehydrogenase (LDH) were used as model proteins. All of the three proteins exhibited a concentration-dependent loss of activity after freezing, with significantly higher recovery at higher concentration. The freezing method and the type of buffer were also important, with sodium phosphate buffer and freezing by immersion of vials in liquid nitrogen associated with the lowest recovery of activity. Differential scanning calorimetry was predictive of the onset of collapse during freeze-drying only for beta-galactosidase. For the other proteins, either no Tg' transition was observed, or the apparent glass transition did not correlate with the microscopically-observed collapse temperature. The time course of activity loss for beta-galactosidase and LDH was compared during freeze-drying under conditions which produced collapse of the dried matrix and conditions which produced retention of microstructure in the dried solid. Recovery of activity decreased continuously during primary drying, with no sharp drop in recovery of activity associated with the onset of collapse. The most important drying process variable affecting recovery of activity was residual moisture level, with a dramatic drop in activity recovery associated with residual moisture levels less than about 10%.

Freeze-drying of tert-butyl alcohol/water cosolvent systems: effects of formulation and process variables on residual solvents

The objective of this study was to identify significant formulation and processing variables affecting levels of tert-butyl alcohol (TBA) and isopropyl alcohol (IPA) in freeze-dried solids prepared from TBA/water cosolvent systems. The variables examined were the physical state of the solute (crystalline vs amorphous), initial TBA concentration, freezing rate, cake thickness, and the temperature and duration of secondary drying. Sucrose and glycine were used as models for noncrystallizing and crystallizing solutes, respectively. The TBA concentration above which eutectic crystallization takes place was determined by differential scanning calorimetry. Model formulations were subjected to extremes of freezing rate by either dipping in liquid nitrogen or by slowly freezing on the shelf of a freeze-dryer. Dynamics of solvent loss during secondary drying was determined by withdrawing samples as a function of time at different shelf temperatures using a thief system. On the basis of these studies, the most important determinant of residual TBA level is the physical state of the solute. Freeze-dried glycine contained very low levels of residual TBA (0.01-0.03%) regardless of freezing rate or initial TBA concentration. For freeze-dried sucrose, residual TBA levels were approximately 2 orders of magnitude higher and were significantly affected by initial TBA concentration and freezing rate. For the sucrose/TBA/water system, relatively low residual TBA levels were obtained when the initial TBA level was above the threshold concentration for eutectic crystallization of TBA, whereas samples freeze-dried from solutions containing TBA concentrations below this threshold contained significantly higher levels of TBA. Residual IPA levels increased continuously with initial concentration of TBA in the sucrose/TBA/water system. Formulations of sucrose/TBA/water which were frozen rapidly contained residual TBA levels which were approximately twice those measured in the same formulation after slow freezing and drying under the same conditions. For the sucrose/TBA/water system, the temperature and time of secondary drying had only minimal influence on residual TBA in the freeze-dried solid. At low initial TBA concentrations (2%), residual TBA increases with increased cake thickness, perhaps because of the influence of depth of fill on effective freezing rate.

Kinetics of Low pH Induced Aggregation of Equine IgG

The kinetics of soluble aggregate formation in equine IgG was studied in the pH 3.4-4.3 range and ionic strength between 0.02 and 0.5 M, and a diagram describing aggregation kinetics as diffusion limited, reaction limited, or transitional as a function of pH and ionic strength was constructed. Aggregation rate is limited by the degree of electrostatic repulsion between the protein molecules in the pH 4.0-4.5 range. Below pH 4.0, a greater degree of attractive force is present, most likely from protein unfolding, and electrostatic repulsion no longer determines the rate of aggregation. The aggregation rate increases with decreasing pH, and at pH 3.4 the aggregation rate is diffusion limited. The pH range separating reaction-limited and diffusion-limited kinetics decreases with increasing ionic strength, indicating charge shielding from the buffer solution influences the aggregation rate. Copyright 1997 Academic Press.

Evaluation of manometric temperature measurement as a method of monitoring product temperature during lyophilization

The objective of this study was to evaluate manometric temperature measurement as a non-invasive method of monitoring product temperature during the primary drying phase of lyophilization. This method is based on analysis of the transient response of the chamber pressure when the flow of water vapor from the chamber to the condenser is momentarily interrupted. Manometric temperature measurements (MTM) were compared to product temperature data measured by thermocouples during the lyophilization of water, mannitol, lactose and potassium chloride solutions. The transient pressure response was mathematically modeled by assuming that four mechanisms contribute to the pressure rise: 1) direct sublimation of ice through the dried product layer at a constant temperature, 2) an increase in the temperature at the sublimation interface due to equilibration of the temperature gradient across the frozen layer, 3) an increase in the ice temperature due to continued heating of the frozen matrix during the measurement, and 4) leaks in the chamber. Experimental transient pressure response data were fitted to an equation consisting of the sum of these terms containing three variables corresponding to the vapor pressure of ice, product resistance to vapor flow, and the vial heat transfer coefficient. Excellent fit between the mathematical model and the experimental data was observed, and the value of the variables was calculated from the measured transient pressure response by a least squares method. The product temperature measured by MTM, which measures the temperature at the sublimation interface, was compared with product temperature measured by thermocouples placed in the bottom center of the vials. Manometrically measured temperatures were consistently lower than the thermocouple measurements by about 2 degrees C, this difference being largely accounted for by the temperature gradient across the frozen layer. The resistance of the dried product to mass transfer calculated from MTM was found to agree reasonably well with values measured by a direct vial technique. Product resistance was observed to increase with increasing solute concentration, and to increase continuously as the depth of the dried product layer increases for mannitol and potassium chloride. For lactose, product resistance increases continuously with thickness up to the onset of collapse, at which point the product resistance becomes essentially independent of depth. Scanning electron microscopy was used to explain this observation based on changes in morphology of the solid. The vial heat transfer coefficients obtained from regression analysis were on the order of 10(-3)-10(-4) cal.sec-1. degrees C-1; however, the scatter in the vial heat transfer coefficient data prevents the method from being used for accurate measurement of the vial heat transfer coefficient. The results of the study show that the manometric method shows promise as a process development tool and as an alternative method of in-process product temperature measurement during primary drying.

Role of the electrostatic attractive force in the adsorption of proteins by aluminum hydroxide adjuvant

The fact that both aluminum hydroxide adjuvant and proteins have a pH dependent surface charge means that electrostatic forces play a role in the adsorption of proteins by aluminum hydroxide adjuvant during the preparation of vaccines. The objective of this study was to examine the contribution of the electrostatic attractive force in the adsorption of proteins by aluminum hydroxide adjuvant. Since the surface charge characteristics of aluminum hydroxide adjuvant can be modified by the adsorption of phosphate anion, a series of aluminum hydroxide adjuvants were prepared by treatment with various concentrations of phosphate anion. The isoelectric points (iep) of these adjuvants ranged from 11.0 to 4.6 and the electrophoretic mobilities at pH 7.4 ranged from 2.0 to -3.3 microns cm/V s. The line broadening of the (020) band of the X-ray diffraction pattern indicated that treatment with phosphate anion did not change the primary crystallite dimension. Adsorption at pH 7.4 of positively charged lysozyme (iep = 11.1) was directly related to the negative surface charge of the adjuvant. No adsorption occurred when the surface charge was positive. In contrast, negatively charged ovalbumin (iep = 4.6) was adsorbed by all of the adjuvants at pH 7.4, although the adsorptive capacity was the greatest when the surface charge was positive. The results indicate that adsorptive forces in addition to the electrostatic attractive force play an important role in the adsorption of some proteins by aluminum hydroxide adjuvant. It is believed the structurally flexible proteins, like ovalbumin, exhibit more complex adsorption behavior than structurally rigid proteins, like lysozyme, for which adsorptive behavior can be explained by electrostatic forces.

The physical state of nafcillin sodium in frozen aqueous solutions and freeze-dried powders

The purpose of this study was to develop a better understanding of the physical chemistry of freeze drying of lyotropic liquid crystals using nafcillin sodium as a model solute. Solutions and freeze-dried powders of nafcillin sodium were studied by polarized light microscopy, differential scanning calorimetry, x-ray powder diffraction, and water vapor adsorption. Differential scanning calorimetry thermograms of nafcillin sodium solutions contain a melting endotherm at approximately -5.5 degrees C and, depending on the concentration and heating rate, a crystallization exotherm immediately after this endotherm followed by the melting endotherm of ice. When the sample is annealed at -4 degrees C, both the endotherm and exotherm are eliminated, and a new endotherm appears at approximately -1 degree C on the shoulder of the ice-melting endotherm. The data are interpreted as melting of a liquid crystalline phase, followed by crystallization. X-ray powder diffractograms of unannealed freeze-dried nafcillin sodium are consistent with a lamellar liquid crystal. Diffractograms of annealed freeze-dried nafcillin sodium indicate crystalline material which is a different crystal form than the monohydrate starting material. Moisture adsorption isotherms of the freeze-dried annealed (crystalline) and unannealed (liquid crystalline) nafcillin sodium show different affinities for moisture compared to the crystalline starting material. Solid-state stability data demonstrate that the freeze-dried liquid crystalline form of nafcillin sodium is much less stable than the freeze-dried crystal-line material. The literature recognizes two types of solute behavior on freezing, where the solute either crystallizes from the freeze concentrate or remains amorphous. Lyotropic liquid crystal formation during freezing represents a separate category of freezing behavior, the physical chemistry of which is worthy of further investigation.

Design and application of a low-temperature Peltier-cooling microscope stage

A light microscopy system has been designed for freezing and lyophilization studies of protein pharmaceuticals. The system consists of a cascade of four Peltier thermoelectric modules in the lyophilization cell to freeze samples to -60 degrees C, controllers to regulate temperature and pressure conditions, and a video camera to record the events under study. Specific demonstration of the system was conducted using recombinant CD4-IgG and human growth hormone (hGH) as model proteins. Observations of recrystallization during warming of frozen CD4-IgG solution and lyophilization of hGH solution are discussed. These examples demonstrate that the system is a useful tool for the fundamental understanding of freezing and lyophilization of protein pharmaceuticals.

Glycine crystallization during freezing: the effects of salt form, pH, and ionic strength

PURPOSE

The purpose of the study is to characterize glycine crystallization during freezing of aqueous solutions as a function of the glycine salt form (i.e., neutral glycine, glycine hydrochloride, and sodium glycinate), pH, and ionic strength.

METHODS

Crystallization was studied by thermal analysis, microscopy, x-ray diffraction, and pulsed Fourier transform nmr spectroscopy.

RESULTS

A solution of neutral glycine with no additives undergoes rapid secondary crystallization during freezing, forming the beta polymorph, with a eutectic melting temperature of -3.4 degrees C. Glycine hydrochloride solutions undergo secondary crystallization relatively slowly, and the eutectic melting temperature is -28 degrees C. Sodium glycinate crystallizes from frozen solution at an intermediate rate, forming a eutectic mixture with a melting temperature of -17.8 degrees C. Where secondary crystallization does not occur rapidly, a complex glass transition is observed in the -70 degrees to -85 degrees C temperature range in the DSC thermograms of all systems studied. Rates of secondary crystallization and the type of crystal formed are influenced by solution pH relative the the pKs of glycine, and also by the change in ionic strength caused by adjustment of pH. Increased ionic strength significantly slows the crystallization of neutral glycine and promotes formation of the gamma polymorph. Thermal treatment or extended holding times during the freezing process may be necessary in order to promote secondary crystallization and prevent collapse during freeze drying.

CONCLUSIONS

The results underscore the importance of recognizing that seemingly minor changes in formulation conditions can have profound effects on the physical chemistry of freezing and freeze drying.