Enabling Freeze-Thaw Stability of PBS-Based Formulations of a Monoclonal Antibody Part III: Effect of Glass Transition Temperature

Abstract

Phosphate buffered saline (PBS) is a physiologically suitable formulation vehicle often used for biologics in research and preclinical studies, but it is notorious for freeze-thaw instability. Here, the authors report differential scanning calorimetry (DSC) analysis of PBS-based formulations of a monoclonal antibody (mAb) designed to prevent protein aggregation upon frozen storage. Flash freezing at -90 °C followed by slow heating at 2 °C/min drastically increased the sensitivity of the method and enabled detection of a transition temperature (Tg*) occurring immediately after the glass transition temperature (Tg’). A Tg* as low as -73.6 °C was detected for the mAb solution in PBS, which decreased by 25 °C with increase in protein concentration from 5 to 20 mg/mL. Lowering the concentration of sodium chloride (NaCl) to 30 mM dramatically increased Tg* by 40 °C. Addition of polyols also resulted in a marked increase in Tg* value, likely due to blending of miscible excipients. To the authors’ knowledge, the current study is the first report on DSC of a mAb formulated in PBS to demonstrate an effective means of increasing glass transition temperature, making frozen storage of mAbs in PBS-based formulations kinetically stable at temperatures of practical use. The results provide a mechanistic understanding of the behavior of PBS-based compositions on freeze-thaw stress and long-term frozen storage (as reported in Part I and Part II of this series, respectively).  

One of the essential elements in biotherapeutic formulation development is optimizing stability against freeze-thaw stress for long-term storage of bulk drug substance and reference standard (1). Protein reagents are also often stored frozen in PBS (2). While frozen storage can significantly extend shelf life by halting chemical reactions, it is also known for adverse impact on the physical stability of proteins, manifested primarily as aggregation upon thawing (1,3,4,5). Because of its similarity to physiological fluids, phosphate buffered saline (PBS) is a common solution in which proteins are formulated for studying the protein’s effect on cells or in animals. PBS, however, is particularly notorious for freeze-thaw instability of proteins (6). An approach to stabilize PBS-based formulations could thus provide a simple physiological solution for use of proteins in research, preclinical, diagnostics, and clinical studies, as well as commercial biotherapeutic products.

Various factors contribute to freeze-thaw instability, often related to the steep temperature dependence of phosphate pKa [the acid dissociation constant], resulting in a significant drop in pH upon freezing (7), as well as the high propensity of sodium phosphate dibasic crystal formation (8). While various factors contribute to instability of frozen protein formulations, kinetic stability of a frozen formulation can often be achieved by storing at temperatures at least 20 °C below the glass transition temperature (Tg’), where the loss of molecular mobility prevents potential degradation (3). Differential scanning calorimetry (DSC) has long been a valuable tool for determining Tg’, although it is often challenging to determine Tg’ for protein solutions (9). Multiple studies have reported Tg’ values of various crystallizing and amorphous agents, some of which are routinely used as excipients in biotherapeutic formulation development (10-15 ).

Here, the authors report DSC analysis of a monoclonal antibody (mAb) formulated in PBS-based formulations to achieve mechanistic understanding of the stabilizing and destabilizing effects observed by freeze-thaw stress testing and long-term frozen storage of the mAb, which were reported in Parts I and II of this series, respectively (16, 17). Flash freezing at -90 °C, followed by slow heating at 2 °C/min, was used to enhance the sensitivity of DSC to detect the onset of crystallization of water and excipients trapped in an amorphous phase during quench freezing. The approach was applied to evaluate the effects of sodium chloride (NaCl), protein concentration, and polyol addition on Tg’ of a mAb formulated in PBS. To the authors’ knowledge, the current study is the first systematic report on the DSC analysis of PBS-based mAb formulations to identify factors that increase thermal phase transition values. Understanding these factors can help facilitate the design of thermodynamically stable frozen mAb in PBS-based formulations at temperatures of practical importance to biotherapeutics.

Materials and methods

Recombinant human mAb was manufactured at Morphotek using a routine platform mAb production process. PBS containing 10 mM sodium phosphate, 150 mM NaCl, pH 7.2 was also prepared at Morphotek.

Sodium phosphate monobasic, monohydrate (United States Pharmacopeia [USP] grade), sodium phosphate dibasic heptahydrate (USP grade), and sucrose crystal (multi-compendial, National Formulary [NF] grade) were obtained from JTBaker. Sodium chloride (USP grade) was sourced from DPH. Sorbitol, Emprove (European Pharmacopoeia [Ph. Eur.], British Pharmacopoeia [BP], Japanese Pharmacopoeia
[JP], NF grade) was purchased from EMD Millipore.

DSC experiments were performed on a TA Instruments Q2000 DSC with nitrogen purge gas. A refrigerated cooling accessory enabled operating to temperatures as low as -90 °C. Indium metal standard was used for single-point temperature and heat flow calibration. Sample sizes of 20 ± 2 mg were sealed in Tzero hermetic aluminum pan (TA Instruments) and manually loaded into the DSC cell that had been pre-chilled to -90 °C. After five minutes of equilibration, the sample was heated at a rate of 2 °C/min. Analysis of the heating data shows exothermic peaks for crystallization of water and/or excipients trapped in the amorphous form during flash-freezing. The peak of this exotherm (Tg*) was obtained and used for correlation with the long-term storage (described in Part II of this series [17]). These crystallization processes occur slightly above Tg’ due to slow heating, and therefore, indicate that the measured temperature of the exothermic peak Tg* is a reasonable approximation of the Tg’.

Results

Rationale for selection of the DSC conditionsTemperature change conditions: In the current study, quench freezing followed by slow heating was used for more sensitive and accurate detection of the temperature(s) where increased molecular mobility occurs (i.e., Tg’) in the amorphous components of flash-frozen samples. The DSC procedure applied in the current study is similar to those described in earlier reports on aqueous solutions of polyols and polymers (10, 14).

Because Tg’ is typically a broad and very weak, low-energy transition, it may not be possible to measure it directly in protein samples at mg/mL concentrations. However, crystallization of amorphous material is a much more energetic process that should have heat flow rates measurable by DSC once the temperature is above Tg’. The presence of a crystallization peak, therefore, indicates that the measured temperature is above Tg’. In this paper, the term Tg* is used to denote that the glass transition temperature of the frozen solution (Tg’) has been exceeded, as measured by the peak of a crystallization process. Tg* values are reported as midpoints of the detected water and excipient crystallization peaks during slow heating of quenched frozen solutions, as they are more reproducibly and accurately detected than the temperature of onset of these peaks.

Pan type: Hermetic pans are typically used for pharmaceutical materials, especially liquids. These pans are sealed, preventing the loss of volatile components and the appearance of the relatively large endothermic peak associated with their evaporation. 

Sample amount: A sample amount of 20 mg was used, which is larger than the amount (2-5 mg) typically used for studying pharmaceutical solids. Although the use of a higher volume DSC pan makes glass transition broader, it was necessary to achieve increased sensitivity to detect and quantify heats of transition for weak transitions such as those observed in this study. DSC instruments measure the rate of heat flow, which is proportional to the sample amount (heat flow = W/g, where W is Joules heat/sec and g is sample amount in grams). The disadvantage of a larger sample amount is a reduction in transition resolution given that the transition occurs over a wider temperature range due to thermal lag. Larger sample sizes are thus recommended only when detection of subtle transitions is the objective, but the experiment needs to be set up in a way that enables detection of such transitions, as described in the following section.

Heating rate: A slower heating rate of 2 °C/min, as compared to that (5-10 °C/min) used for most applications of DSC was selected. There were two reasons for using the slower-than-average rate in this study:

• Given the larger-than-typical sample amount used for increasing sensitivity, the loss in transition resolution is minimized by using a slower heating rate as it provides more time for transitions due to structural changes to occur.

• Although heating rate does not significantly affect the onset temperature of thermodynamic changes in structure (such as melting), the onset temperature of kinetic processes always increases as heating rate is increased due to the activation energy of the process. The temperature of crystallization of unfrozen water in the frozen solution (a kinetic process) as measured in the DSC at slow heating rates is closer to that occurring under actual storage conditions.

Thermal history: Quench cooling, significantly faster than rates routinely used in DSC studies, is achieved by manually loading the sample pan into the DSC cell that has been pre-cooled to -90 °C. Such quench cooling rapidly decreases molecular mobility, trapping non-crystallized excipients and water that would normally crystallize on freezing. As this frozen matter is heated, the trapped non-crystallized components start crystallizing because of the increased molecular mobility when reaching above the Tg’ of formulation components. As shown in the following section, this leads to crystallization of unfrozen water/excipients upon heating of the sample with a relatively large exothermic crystallization peak, enabling the detection of the temperature where mobility increases in the amorphous structure. As described previously, a slow heating rate of 2 °C/min enables detection of the temperature of increased mobility (Tg*) with a higher precision than the faster heating rates routinely used in pharmaceutical studies (10,14).

Thermal transitions of mAb formulated in PBS 

A 5 mg/mL mAb solution in PBS showed a broad endothermic transition at -73.6 °C (see Figure 1A). This transition is believed to be due to crystallization of unfrozen water upon heating above Tg’, as discussed in the previous section. Figure 2 demonstrates that annealing of the frozen mAb sample at temperatures above Tg* (-30 °C) and below melting temperature (Tm) eliminates the crystallization peak on re-heating, because water and excipients crystallize during annealing.

Figure 1: The effect of monoclonal antibody (mAb) concentration on glass transition temperature by differential scanning calorimetry (DSC). A. mAb formulated in phosphate buffered saline (PBS) containing 150 mM sodium chloride (NaCl).

In a frozen solution held above Tg*, a concentrated amorphous solution of protein and formulation excipients is formed as a “freeze-concentrate” (3), in which there is likely adequate protein mobility to allow protein aggregation. Thus, freezing and storing of the mAb (5 mg/mL in PBS) at -70 °C was unstable (unpublished results). While storage at -80 °C could be better than at -70 °C, the 6 °C difference from Tg* of -73.6 °C is still not at the desirable 20 °C difference to ensure low mobility of the amorphous phase and does not render a kinetically stable frozen formulation.

Figure 1: The effect of monoclonal antibody (mAb) concentration on glass transition temperature by differential scanning calorimetry (DSC). B. mAb formulated in PBS containing 75 mM NaCl.

The exothermic phase transitions at approximately -20 °C were also observed (Figure 1A-C). The major peak was most likely due to NaCl crystallization. Hence, storage at -20 °C was found to be thermodynamically unstable and may result in a physically, and thus, potentially chemically unstable condition for this mAb in PBS. The data are similar to the findings of mAb instability on frozen storage described in the literature (18).

Figure 1: The effect of monoclonal antibody (mAb) concentration on glass transition temperature by differential scanning calorimetry (DSC). C. mAb formulated in PBS containing 30 mM NaCl.

Effect of protein and salt concentration: Increasing protein concentration from 
5 mg/mL to 20 mg/mL shifted Tg* to -48.9 °C (see Figure 1B), consistent with greater stabilization against aggregation at higher protein concentration during long-term storage (17). Likewise, decreasing the salt concentration shifted the glass transition temperature to even higher temperatures such that at 1 mg/mL protein and 30 mM NaCl, Tg* was -33.3 °C (see Figure 1C); this observation renders -80 °C a physically stable storage temperature for the modified formulation. With decreasing salt concentration from 150 mM to 30 mM, the heat flow of the phase transition near -25 °C, calculated from area under the peak, decreases by about eight-fold, from 6 J/g (see Figure 1A) to 0.8 J/g (see Figure 1C), further evidence that the endothermic transition is related to NaCl.

Figure 2: Annealing of monoclonal antibody (mAb) formulated at 10 mg/mL in phosphate buffered saline (PBS) containing 75 mM NaCl.

Effect of polyols: At a constant protein concentration (5 mg/mL), the addition of polyols removes the NaCl crystallization transition and increases Tg* from -73.6 °C to -59.2 °C for 3% sorbitol and  -48.7 °C for 5% sucrose (see Figure 3). The effect is likely because of the blending of sucrose or sorbitol with the salts in PBS, analogous to the phenomena described in the literature (10-12, 19-20). Sucrose should provide a more rugged formulation for freezing at -70 °C than sorbitol because of the observed 10 °C higher Tg* (-48.7 °C vs -59.2 °C), although both polyols increased Tg* sufficiently for thermodynamically favorable storage at -80 °C. 

Figure 3: The effect of polyol on glass transition temperature of monoclonal antibody (mAb) formulated in phosphate buffered saline (PBS) containing 150 mM sodium chloride (NaCl).

The Tg* values of the mAb formulations analyzed in the current study are summarized in Table I. The compiled data show the observed trends for the effects of protein and NaCl concentrations, as well as the addition of polyols, on Tg*. While the accuracy of the Tg* values in this study is determined by the applied DSC methodology for weak transition detection and the inherently small DSC sample size that prevents evaluation of kinetic factors affecting the glass transitions, the data represent a valuable estimate of the Tg’, reflecting on molecular mobility in various frozen PBS-based mAb formulations.

Table I: Glass transition temperature Tg* values of a monoclonal antibody (mAb) formulated in phosphate buffered saline (PBS) at varying protein and sodium chloride (NaCl) concentrations.

Discussion

Developing a long-term, stable, frozen protein formulation requires an understanding of both the thermodynamics of the frozen state and the kinetics of events occurring during the freezing and thawing process. Feasibility studies should demonstrate stability of the formulation to freezing and thawing, which are often conducted by multiple freeze-thaw cycling stress tests. These studies should also provide an understanding of the molecular events that occur in the frozen state. The current study provides a systematic approach for an analysis of the effect of formulation composition of the PBS-based vehicles on Tg*. This approach is recommended as a predictive tool for molecular events that potentially affect long-term frozen storage stability. The DSC method used in this study predicted that long-term storage at -20 °C or -70 °C could be problematic for PBS formulations of the mAb. The trends in destabilizing and stabilizing the mAb on freeze-thaw in the PBS-based formulations are consistent with the current DSC study findings (16). Long-term stability data on the formulations listed in Table I under different freezing temperature conditions were found to be consistent with the observed Tg* of those formulations (17). A significant increase in Tg* upon polyol addition illustrates an approach to achieving thermodynamically stable, frozen PBS-based formulations.

This study also revealed a dramatic effect of low salt concentrations on Tg* of the PBS-based formulations (see Table I). Lowering the NaCl concentration from 150 mM to 30 mM NaCl increased Tg* by more than 40 °C (from the extremely low value of -75 °C to -32°C). This impact was irrespective of protein concentration (1-10 mg/mL). On the other hand, protein concentration had a large effect on Tg* at 150 mM NaCl, decreasing Tg* by 25 °C when increasing protein from 5 mg/mL to 20 mg/mL. These DSC findings are consistent with the stabilization observed at higher protein and lower salt concentrations following a single freeze-thaw (16). The destabilizing effect observed for low salt/low protein concentration when stored at -20 °C (17) is also consistent with the DSC finding of a Tg* that is only 12 degrees higher than the storage temperature. Further studies are required for mechanistic understanding of the observed NaCl effect on material properties of the frozen PBS-based formulations.

Conclusion

Although the current approach does not enable direct measurement of Tg’, it allows detection of extremely weak glass transitions that are usually not amenable to DSC and modulated DSC detection. In contrast to the published studies aimed at detection of weak glass transitions using >5 °C/min heating rates, the slow heating performed at the 2 °C/min rate allows more accurate detection of temperatures of increased mobility, called Tg* in the current study. The measured Tg* could be different from the true Tg’ value, but it could not be lower than Tg’. Thus, the extremely low Tg* transition temperature unexpectedly detected in a PBS-formulated mAb solution presents the first such report in literature to the best of the authors’ knowledge. In addition, the effect of polyols on Tg* in the PBS-formulated mAb represents, to the authors’ knowledge, the first report of the blending of such well-recognized and employed excipients for biotherapeutic formulations. The current study provides a basis for a feasible introduction of PBS-based formulations into the biotherapeutics practice.

References

1. J. Patel et al., Bioprocess Tech. January: 20-31 (2011).
2. Abcam antibody storage guide.
3. B. S, Bhatnagar, R. H. Bogner, M. J. Pikal, Pharm. Dev. Technol. 12, 505-523 (2007).
4. W. Wang, Intern. J. Pharm. 289, 1-30 (2005).
5. L. A. Kueltzo et al., J. Pharm. Sci. 97 (5), 1801-1812 (2008).
6. K. A. Pikal-Cleland et al., Biochem. Biophys. 384 (2), 398-406 (2000).
7. G. Gomez, M. J. Pikal, N. Rodriguez-Hornedo, Pharm. Res. 18, 90-97 (2001). 
8. A. Pyne, K. Chatterjee, R. Suryanarayanan, Pharm. Res. 20 (5) 802-803 (2003).
9. Sh. Jiang, S. L. Nail, Eur. J. Pharm. Biopharm. 45, 249-257 (1998). 
10. L.-M. Her, M. Deras, S. L. Nail, Pharm. Sci. 12 (5) 768-772 (1995).
11. K. Izutsu et al., Pharm. Res. 13 (9) 1393-1400 (1996).
12. K. I. Izutsu et al., Eur. J. Pharm. Biopharm. S0939-6411 (2013). 
13. K. Izutsu, Sh. Kojima, Pharm. Res. 17 (10), 1316-1322 (2000).
14. Y. H. Roos, M. Karel, Cryoletters. 12 (6) 367-376 (1991).
15. H. D. Golf, Food Res. Intern. 25, 317-325 (1992).
16. T. Mezhebovsky et al., Biopharm Inter., 26 (8) 33-39 (2016).
17. T. Mezhebovsky et al., BioPharm Inter., 29 (9) 38-47 (2016).
18. S. Singh et al., Pharm. Res. 28, 873-885 (2011).
19. W. Brostow et al., Materials Letters. 62, 3152-3155 (2008).
20. D. M. Piedmonte et al., Pharm. Res. 24 (1) 136-146 (2007).

Article Details

BioPharm International
Vol. 29, No. 10
Page: 31–36

Citation

When referring to this article, please cite as L. Thomas et al., "Enabling Freeze-Thaw Stability of PBS-Based Formulations of a Monoclonal Antibody Part III: Effect of Glass Transition Temperature," BioPharm International 29 (10) 2016.