- Email: [email protected]
Effect of Freezing on Lyophilization Process Performance and Drug Product Cake Appearance
Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
Contents lists available at ScienceDirect
Journal of Pharmaceutical Sciences journal homepage: www.jpharmsci.org
Pharmaceutical Biotechnology
Effect of Freezing on Lyophilization Process Performance and Drug Product Cake Appearance Reza Esfandiary, Shravan K. Gattu, John M. Stewart, Sajal M. Patel* Department of Formulation Sciences, Biopharmaceutical Development, MedImmune, Gaithersburg, Maryland 20878
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 November 2015 Revised 8 January 2016 Accepted 2 February 2016
This study highlights the significance of the freezing step and the critical role it can play in modulating process performance and product quality during freeze-drying. For the model protein formulation evaluated, the mechanism of freezing had a significant impact on cake appearance, a potential critical product quality attribute for a lyophilized drug product. Contrary to common knowledge, a freezing step with annealing resulted in 20% increase in primary drying time compared to without annealing. In addition, annealing resulted in poor cake appearance with shrinkage, cracks, and formation of a distinct skin at the top surface of the cake. Finally, higher product resistance (7.5 cm2.Torr.hr/g) was observed in the case of annealing compared to when annealing was not included (5 cm2.Torr.hr/g), which explains the longer primary drying time due to reduced sublimation rates. An alternative freezing option using controlled ice nucleation resulted in reduced primary drying time (i.e., 30% reduction compared to annealing) and a more homogenous batch with elegant uniform (i.e., significantly improved) cake appearance. Here, a mechanistic understanding of the distinct differences in cake appearance as a function of freezing mechanism is proposed within the context of ice nucleation temperature, ice crystal growth, and presumed solute distribution within the frozen matrix. © 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
Keywords: freeze-drying/lyophilization drying glass transition solid-state stability porosity processing process analytical technology (PAT) protein formulation
Introduction Freeze-drying is a commonly used unit operation to achieve acceptable shelf life for an otherwise unstable molecule. Freezedrying is usually carried out in three steps: freezing (i.e., conversion of water into ice), primary drying (i.e., removal of ice by sublimation at low temperature and pressure), and secondary drying (i.e., removal of unfrozen water). Typically, water content at the end of the secondary drying step is 1%. The ice nucleation temperature (or degree of supercooling) is a potential critical process parameter during the freezing step.1 The pore structure in the dried matrix is a reflectance of how the solution was frozen with every 1 C increase in ice nucleation temperature resulting in a 1%-3% reduction in primary drying time.2 A higher degree of supercooling results in smaller ice crystals, which results in higher product resistance and longer primary drying time but shorter secondary drying (due to higher specific surface area [SSA]). Thus, the freezing step affects not only primary drying but also secondary drying. The freezing step * Correspondence to: Sajal M. Patel (Telephone: þ1-301-398-5247; Fax: þ1-301398-7782). E-mail address: [email protected] (S.M. Patel).
of the freeze-drying process has received significant attention in the last decade because this first step of the process governs not only the performance of the subsequent steps (i.e., primary and secondary drying) but also the product quality (i.e., physical stability, cake appearance, residual moisture, reconstitution time).3 The ice nucleation temperature, random or stochastic in nature, is also an important process development and scale-up issue. Vials within a batch may vary in ice nucleation temperature from as high as 2 C to as low as 18 C. Thus, there is heterogeneity in ice nucleation temperature not only within a batch but also from batch to batch. An additional consideration is the clean room (Class 100) environment in manufacturing compared to laboratory bench preparation for a lab-scale dryer, resulting in differences in ice nucleation temperature.2 Annealing during freezing is often a common approach to mitigate heterogeneity in ice nucleation temperature, reduce primary drying time,3 and crystallize any crystallizing excipients in the formulation.4 During annealing, product temperature is maintained, for a certain period of time, above the glass transition temperature (Tg’) but below the melting temperature of the formulation matrix, resulting in Ostwald ripening (i.e., growth of bigger ice crystals at the expense of smaller ones). Although optimization of annealing time and temperature is critical, annealing
http://dx.doi.org/10.1016/j.xphs.2016.02.003 0022-3549/© 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
1428
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
may not always be suitable from a product quality standpoint due to conformational changes in protein structure or amorphousamorphous phase separation.5,6 In addition, for formulation matrices wherein crystallization of solute is involved, annealing may not always result in shorter primary drying time (i.e., lower dry layer resistance).7 Nonetheless, annealing during freezing is a way to mitigate the heterogeneity in ice nucleation temperature. Other alternative options have become available in the recent past to control ice nucleation temperature.8-11 The effect of ice nucleation temperature on process performance (i.e., primary drying time) using controlled ice nucleation techniques is well established.1,3,12,13 The SSA of the dried matrix is a good surrogate of the initial freezing characteristics of a solution. Higher ice nucleation temperature results in lower SSA and hence shorter primary drying time. Developing an understanding of the effect of controlled ice nucleation on product quality is now becoming critical to enable controlled ice nucleation techniques as process analytical tools to control ice nucleation temperature in clinical and commercial batches. This work, presented as an industry case study, assesses the negative effect of annealing on cake appearance and demonstrates how controlled ice nucleation results in significant improvement in cake appearance as well as batch homogeneity. Cake appearance is a potential critical product quality attribute with a specification both for lot release and stability. The finished product is visually inspected (100%) for cake appearance because a change or variability in cake appearance from what is anticipated is likely an indication of change in product quality. Although the acceptance criterion for cake appearance is outside the scope of this work, the study here clearly demonstrates the significance of the freezing step and its impact on cake appearance.
(SP Scientific, Stone Ridge, NY) at 20% of total lyophilizer shelf load. A 15-min equilibration at 5 C was used before initiation of the freeze-drying cycle. Although different freezing methods were used (Table 1), the primary drying (30 C, 100 mTorr) and secondary drying (40 C, 100 mTorr) steps were kept identical for all the cycles tested. The pressure was controlled using a capacitance manometer gauge. The duration of the secondary drying step was fixed at 6 h in all the freeze-drying cycles whereas the duration/end of the primary drying step was determined by a comparative pressure measurement wherein the Pirani gauge pressure measurement converges, within 10 mT, with that of the capacitance manometer chamber pressure.14,15 The product temperature during freeze-drying was monitored using thermocouples (TCs) and also determined via Manometric temperature measurement (MTM). A total of 4 TCs were placed at the bottom center of selected vials located at the edge (2 vials) and center (2 vials) of each shelf. The TCs were distributed across the shelf such that any variation in product temperature as a function of shelf location was monitored. To evaluate the impact of the freezing step on overall process efficiency and product quality, the freeze-drying cycles were designed to differ in the freezing step. One of the freeze-drying cycles included an annealing step with a temperature set point of 15 C (referred to as Anneal), whereas the other cycle used the ControLyo™ Technology10 for a controlled ice nucleation step (referred as CN) at a temperature set point of 8 C in place of annealing. Finally, a third cycle without either annealing or controlled ice nucleation was evaluated (referred to as No Anneal). The freeze-drying process parameters for all three cycles are summarized in Table 1. Determination of Product Resistance and Product Temperature Using MTM Measurement
Materials and Methods Materials The model protein (referred as Protein X) was produced and purified at MedImmune (Gaithersburg, MD). Protein X is a glycoprotein with a size of ~75 kDa, formulated at 10 mg/mL in a citrate-based buffer containing trehalose dihydrate, an arginine/arginine-HCl mixture, and polysorbate 80 (PS80) as excipients. All excipients used were of compendial pharmaceutical grade. Lyophilization Process Parameters For all freeze-drying experiments, 3.5 mL of 0.22-mm filtered Protein X formulation was filled aseptically into 10-mL vials and partially stoppered with 20-mm stoppers (both the vials and stoppers were autoclaved and sufficiently dried before use). The freeze-drying cycles were performed on a Lyostar 3 freeze-dryer
MTM involved closing of the valve connecting the chamber and the condenser for a brief period (~25 s) and recording the pressure rise as a function of time. The pressure rise data were fitted to the MTM equation to determine product temperature at the sublimation interface, product resistance, and vial heat transfer coefficient.16-19 Differential Scanning Calorimetry Differential Scanning Calorimeter Q2000 series from TA instruments (New Castle, DE) was used to determine the glass transition temperature (Tg’). For the Tg’ measurement, 20 mL of liquid sample was added into an aluminum pan and sealed hermetically. An empty pan and lid were used as the reference pan. The sample was frozen to 60 C and then heated to 25 C at a rate of 5 C/min. The Tg’ value was determined using Universal Analysis software (TA instruments, New Castle, DE) and reported as the midpoint of the glass transition.20
Table 1 Lyophilization Process Parameters for the Anneal, No Anneal, and Controlled Ice Nucleation (CN) Cycle Process Parameter/Cycle
Anneal Cycle
No Anneal Cycle
CN Cycle
Ramp to freezing Freezing temperature/hold time Annealing temperature/hold time Controlled ice nucleation temperature/hold time Ramp to primary drying Primary drying temperature/pressure/hold time Ramp to secondary drying Secondary drying temperature/pressure/hold time
1 C/min 40 C/4 h 15 C/2 h NA 0.1 C/min 30 C/100 mTorr/Pirani 0.3 C/min 40 C/100 mTorr/6 h
1 C/min 40 C/4 h NA NA 0.1 C/min 30 C/100 mTorr/Pirani 0.3 C/min 40 C/100 mTorr/6 h
1 C/min 40 C/4 h NA 8 C/4 hr 0.1 C/min 30 C/100 mTorr/Pirani 0.3 C/min 40 C/100 mTorr/6 h
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
1429
Residual Moisture Determination A Mettler Toledo C30 coulometric KF titrator equipped with a Stromboli oven sample changer (Columbus, OH) was used for residual moisture measurements. In brief, the lyophilized cake was broken down using a spatula. About 100 mg of sample was transferred to a clean 20-mL vial and sealed. A blank (i.e., an empty vial) was also prepared and used to equilibrate the system until the drift was below the intended 25-mg water/min baseline value. System suitability was assessed using a reference standard with target 1% residual moisture content. All sample preparations were performed in the glove box with nitrogen purge (< 0.5% relative humidity). Samples were heated for 10 min at 100 C and residual moisture was calculated using the Lab-X software (Mettler Toledo). Scanning Electron Microscopy Scanning electron microscopy (SEM) was used for qualitative assessment of the lyophilized cake morphology, more specifically the pore structure. The lyophilized cake was first cut into large pieces using a clean sharp blade. These pieces were then sliced along the cylindrical axis. The untouched cake surface was then used for SEM examination by a bench top SEM instrument (Model TM-1000, Hitachi, Clarksburg, MD). All samples were sputter coated using gold with 10-nm thickness using a sputter coater from Electron Microscopy Sciences (Model EMS 150R ES, Hatfield, PA). Results Thermal Analysis The glass transition temperature (Tg’) of the maximally freezeconcentrated Protein X formulation was determined using differential scanning calorimetry as detailed in the Methods section. The Tg’ was determined to be about 34 C without any crystallization event (data not shown). Lyophilization Process Data The Protein X formulation was lyophilized using three uniquely designed freeze-drying cycles (Anneal, No Anneal, and CN) with distinct differences in the freezing step. These cycles, as detailed in Table 1, were identical in the primary and secondary drying steps and were run at the same batch size. The impact of the freezing step/mechanism is discussed in the following section with regard to both process performance as well as product quality.
Figure 1. Product temperature determined by MTM during primary drying for the Anneal, No Anneal, and CN cycles.
Primary Drying Time The end of primary drying was defined when the Pirani pressure converged within 10 mTorr of the capacitance manometer. The Pirani pressure profile for the Anneal, No Anneal, and CN cycles is shown in Figure 2. The Anneal cycle resulted in the longest primary drying (about 120 h). The primary drying time was reduced by about 20% and 30% in the No Anneal (about 95 h) and CN cycles (about 85 h), respectively. The significant differences observed in primary drying time are exclusively due to freezing differences because all other parameters (formulation, batch size, vial and stopper type, fill volume, and primary and secondary drying parameters) were kept the same across all the cycles tested. Another interesting observation is the difference in slopes of the Pirani pressure drop with values of about 4, 2.5, and 1.5 mTorr/hr for the Anneal, No Anneal, and CN cycles, respectively. The steepness of the slope of the Pirani pressure is an indication of batch homogeneity, where a steeper slope indicates a more homogeneous batch. Thus, the batch homogeneity follows the trend CN > No Anneal > Anneal.
Process Performance Product Temperature. Product temperature during each freezedrying cycle was monitored using TCs and determined using MTM methodology. The product temperature during primary drying as determined by MTM is compared in Figure 1. During the steady-state stage of the primary drying, the product temperature was similar for No Anneal and CN cycles at about 35.5 C, whereas for the Anneal cycle the product temperature was about 1 C warmer (about 34.5 C). The general trend in product temperature during primary drying followed Anneal > No Anneal > CN. Under all conditions, product temperature was close to or below Tg’ (the collapse temperature, measured using freeze dry microscopy, was within 0.5 C of Tg’, data not shown).
Figure 2. The Pirani chamber pressure during primary drying for the Anneal, No Anneal, and CN cycles. The spikes in the pressure are due to MTM measurement.
1430
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
Product Quality Cake Appearance. Figure 3 shows representative pictures of the lyophilized cakes obtained from the Anneal, No Anneal, and CN cycles with distinct differences observed as a function of the freezing conditions tested. The lyophilized cakes in the Anneal cycle exhibited the poorest quality with extensive cake shrinkage at the bottom of the vials, cracks, and collapse, as well as an enhanced level of porousness and a distinct skin formation at the top of the dried cakes. Samples from the No Anneal cycle, although relatively improved in their appearance compared to the Anneal cycle, still exhibited partial collapse at the bottom of the vial but with no visible skin formation at the top. The vials lyophilized using the CN cycle resulted in elegant cakes with no cake defects. In general, the described cake appearance characteristics within each cycle were consistent across all the vials irrespective of the vial location on the shelf. Product Resistance Product (or dry layer) resistance, a reflection of how the solution is frozen, governs the drying rate (i.e., mass transfer), which in turn governs the product temperature (i.e., heat transfer). To better understand the potential impact of the different freezing methodologies on product resistance, MTM was used to determine product resistance as a function of dry layer thickness. Data showed that vials subjected to annealing (Anneal) exhibited the highest level of product resistance at about 7.5 cm2 Torr h/g while lower values of about 5 and 4 cm2 Torr h/g (product resistance reported at about 0.5 cm of dry layer thickness) were observed for the No Anneal and CN cycles, respectively (Fig. 4). The observed product resistance values correlated well with the primary drying time in Figure 2, with a progressive reduction in product resistance directly correlating with shorter primary drying time. The Anneal cycle with the highest dry layer resistance showed the longest primary drying time of about 120 h, whereas the No Anneal and CN cycles with lower resistance values resulted in shorter primary drying durations of about 95 h and 85 h, respectively. In addition, a unique characteristic of the product resistance data specific to the Anneal
cycle was the significantly high product resistance at the start of drying followed by a decrease in product resistance with increasing dry layer thickness, a characteristic of skin formation followed by micro-collapse/collapse, as supported by the cake appearance (Fig. 3). Scanning Electron Microscopy A qualitative examination of the lyophilized cakes by SEM demonstrated significant differences in pore structure when comparing samples from the Anneal cycle to those from No Anneal and CN cycles. Although the samples, imaged from top cross-sections of the lyophilized cakes in the No Anneal and CN cycles, exhibited similar and perhaps more typical pore structures, those obtained from the Anneal cycle showed a distinct layer at the top, evident by the exhibited solid sheet (i.e., closed pore structure) only observed in case of annealing (Fig. 5). This observation is aligned with cake appearance observations (Fig. 3) where a distinct skin formation was noted for samples from the Anneal cycle. Residual Moisture and Stability of the Lyophilized Cakes The residual moisture content was determined to be 1% across all three cycles. To assess the impact of differences in cake appearance, product from Anneal, No Anneal, and CN cycles was tested for stability at 40 C for 1 month. There was no change in product quality prelyophilization and postlyophilization as well as on storage at 40 C up to 1 month (data not shown), in agreement with previously published work21 that collapse does not affect product stability. Discussion This work clearly highlights the importance of the freezing step in lyophilization of a model protein formulation, impacting both process performance and the product quality. The complex role of ice nucleation and ice crystal growth during freezing is evident from different freezing steps/mechanisms (i.e., annealing, no annealing, and controlled ice nucleation).3
Figure 3. Representative product cake appearance for the Annealed (left column), No Anneal (middle column), and Controlled Nucleation (right column) cycles. The pictures on the top row are captured from the vials positioned inverted whereas the ones on the bottom row are captured from the same vials positioned upright.
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
Figure 4. Product resistance as a function of dry layer thickness determined by MTM for the Anneal, No Anneal, and CN cycles.
Contrary to common knowledge regarding enhanced batch homogeneity and reduced primary drying time with use of annealing, for the particular formulation matrix studied in this work, annealing at 15 C for 120 min resulted in batch heterogeneity and extended primary drying time.7 Data here showed that use of annealing (Anneal) extended the primary drying duration by about 20% compared to No Anneal (Fig. 2). The longer primary drying time in case of annealing can be explained based on the observed higher product resistance (Fig. 4) and skin formation (Figs. 3 and 5) causing reduced sublimation rate. In addition, annealing resulted in skin formation followed by cracks and collapse,22 during primary drying as suggested by the gradual decrease in product resistance values (Fig. 4). Use of controlled ice nucleation proved to be an effective process optimization approach where about 30% reduction in primary drying time was observed compared to the annealed cycle (Fig. 2). In addition, although exclusion of annealing resulted in some improvements in cake appearance with minor cake shrinkage and partial collapse still present, use of the controlled ice nucleation significantly enhanced cake appearance, resulting in elegant lyophilized cakes (Fig. 3). In the case of uncontrolled ice nucleation, the temperature is coldest (i.e., highest degree of supercooling) and the probability of ice nucleation and particulate matter (if any) is highest at the bottom of the vial, in contact with the temperature-controlled
1431
shelves. On ice nucleation, ice crystals grow very rapidly into the supercooled solution. The product temperature then increases (typically up to the equilibrium freezing temperature of the solution) as the heat of crystallization is released from the frozen sample to the cold bottom of the vial. After this initial ice nucleation, the ice crystals grow rapidly toward the surface of the sample resulting in columnar, dendritic, or finger-like ice crystal morphology.23 During annealing, as large ice crystals are formed at the expense of the smaller ice crystals (i.e., Oswald ripening), the highly concentrated solutes trapped between the ice crystals are expelled via a convective flow toward the ice finger tips. Usually the solutes are distributed homogeneously during ice nucleation (Fig. 6, No Anneal cycle); however, they macroscopically redistribute during ice crystal growth from regions of higher concentration toward larger forming ice crystals via convective flow. The ice growth is halted when it reaches the surface of the sample, further resulting in the flow of highly concentrated solutes out of the interstices. This produces a thick layer of concentrated solute on the surface of sample, exhibiting a distinct skin formation after freeze-drying (Figs. 3 and 5, Anneal cycle). This hypothesis is supported by the observation of skin formation on annealing, whereas no skin formation was observed in the absence of the annealing step. The minor collapse observed in the No Anneal cycle could be due to product temperature being close to or exceeding Tg'. In the case of controlled ice nucleation,10 ice first nucleates at the top of the sample followed by a downward propagation, resulting in uniform ice crystal growth and homogeneous distribution of the solutes within the interstices. This is clearly supported by the lower product resistance (i.e., higher porosity) in Figure 4, lower product temperature (Fig. 1), shorter primary drying time (Fig. 2), and elegant cake appearance (Fig. 3). On the contrary, if skin formation were to occur with controlled ice nucleation (as a worstcase scenario), it would be formed at the bottom of the vial (i.e., last to dry), which very likely would go unnoticed during visual inspection because it would have minimal to no impact on the macroscopic cake appearance. This work clearly highlights the significance of the freezing step and the critical role it can play in modulating the process performance24-26 as well as product quality during lyophilization. As detailed in the previous section, the mechanism of freezing has a significant impact on cake appearance, a potential critical product quality attribute for lyophilized products. For the particular case study described here, no impact on stability (i.e., other product quality attributes) was observed as a function of cake appearance (i.e., Anneal, No Anneal, and CN), which is in line with prior publications wherein even a collapsed cake had comparable stability to a noncollapsed cake.21 One can, however, question if mitigating cake appearance defects is critical at the expense of additional resources particularly when stability is not of concern. If one were to accept cake defects, then demonstrating intrabatch and interbatch
Figure 5. Representative scanning electron microscopy (SEM) images of lyophilized cakes from the Annealed (left), No Anneal (middle), and Controlled Nucleation (right) cycles. All images are obtained from the top cross-section of the cakes.
1432
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
Figure 6. Representative schematics of the ice crystal structures for the No Anneal (left), Control Nucleation (middle), and Anneal (right) cycles.
reproducibility still remains critical. Clearly, differences in the freezing step here resulted in significantly different cake appearance, which is a concern because freezing differences do exist between laboratory and production scales, at least in the case of uncontrolled ice nucleation. A clear distinction must be made between poor process control/understanding versus formulation and process limitations to decide whether a cake defect is acceptable. Nonetheless, as mentioned earlier, defining and justifying acceptance criteria for cake appearance are not the focus of this work. In general, to arrive at more widely applied conclusions regarding a potential correlation between cake appearance and physical stability or lack thereof, more comprehensive assessment is needed over a wider formulation and process design space as well as evaluation of real-time, long-term stability under intended storage conditions. The need for a solid-state characterization technique that can help evaluate the subtle differences in cake morphology/pore structure at microscopic level is critical to assess the effect of freezing parameters on finished product quality attributes, specifically cake appearance.27 Although SEM provides visualization of the cake morphology, it is still qualitative and depends on sample preparation. SSA provides a quantitative measurement as measured via the Brunauer-Emmett-Teller technique, but Brunauer-Emmett-Teller is very unlikely to pick up small differences in SSA due to subtle freezing differences. The parafin dye method28 allows visualization of the cake morphology but requires beforehand sample preparation and cake sectioning to view under the microscope. The X-ray micro-CT29 is an emerging technology to evaluate morphology of freeze-dried cakes. A nondestructive technique with the capability to perform quantitative analysis and that can provide visualization of the inside of the cake would be immensely useful to characterize the cake morphology after lyophilization. Another area that needs additional research is the impact of formulation matrices on the freeze-drying process (i.e., why skin formation is observed on annealing with some formulations and not with others). Annealing has been applied during freezing for several formulation and process conditions, and usually no skin formation is observed. However, the presence of protein and the complex interactions within different formulation matrices on freezing needs to be evaluated to better understand the effect of formulation matrix on freezing mechanism. Additionally, the effect of fill volume and vial size (i.e., drug product presentation) needs to be evaluated to identify critical process parameters that govern freezing mechanism. Identifying physical properties of the solution that govern freezing mechanism and establishing correlation with product quality will help build tools that would enable predicting the effect of process parameters on product quality. In future studies, we intend on evaluating formulation and fill-finish parameters using the same model protein to establish the link between formulation and process and its impact on drug product quality.
Conclusion This work clearly demonstrates the effect of the freezing step on lyophilized drug product cake appearance. Annealing, commonly applied during the freezing step, resulted in skin formation, for the conditions evaluated, causing product collapse (at the extreme extent). The design of the freezing step (annealing, no annealing, or controlled ice nucleation) indirectly governs the freezing mechanism, which may further impact product quality (in this case cake appearance). Although primary drying continues to receive the highest attention from process optimization and scale-up perspectives, the criticality of the freezing step in the freeze-drying process design, development, optimization, and scale-up is not to be ignored. Acknowledgments The authors thank Nancy Craighead for the critical review of this manuscript. References 1. Rambhatla S, Ramot R, Bhugra C, Pikal MJ. Heat and mass transfer scale-up issues during freeze drying: II. Control and characterization of the degree of supercooling. AAPS PharmSciTech. 2004;5:e58. 2. Roy ML, Pikal MJ. Process control in freeze drying: determination of the end point of sublimation drying by an electronic moisture sensor. J Parenter Sci Technol. 1989;43:60-66. 3. Searles JA, Carpenter JF, Randolph TW. Annealing to optimize the primary drying rate, reduce freezing-induced drying rate heterogeneity, and determine T(g)' in pharmaceutical lyophilization. J Pharm Sci. 2001;90:872-887. 4. Li X, Nail SL. Kinetics of glycine crystallization during freezing of sucrose/ glycine excipient systems. J Pharm Sci. 2005;94:625-631. 5. Heller MC, Carpenter JF, Randolph TW. Manipulation of lyophilization-induced phase separation: implications for pharmaceutical proteins. Biotechnol Prog. 1997;13:590-596. 6. Lueckel B, Helk B, Bodmer D, Leuenberger H. Effects of formulation and process variables on the aggregation of freeze-dried interleukin-6 (IL-6) after lyophilization and on storage. Pharm Dev Technol. 1998;3:337-346. 7. Lu X, Pikal MJ. Freeze-drying of mannitol-trehalose-sodium chloride-based formulations: the impact of annealing on dry layer resistance to mass transfer and cake structure. Pharm Dev Technol. 2004;9:85-95. 8. Geidobler R, Winter G. Controlled ice nucleation in the field of freeze-drying: fundamentals and technology review. Eur J Pharm Biopharm. 2013;85:214-222. 9. Kasper JC, Winter G, Friess W. Recent advances and further challenges in lyophilization. Eur J Pharm Biopharm. 2013;85:162-169. 10. Konstantinidis AK, Kuu W, Otten L, Nail SL, Sever RR. Controlled nucleation in freeze-drying: effects on pore size in the dried product layer, mass transfer resistance, and primary drying rate. J Pharm Sci. 2011;100:3453-3470. 11. Patel SM, Pikal M. Process analytical technologies (PAT) in freeze-drying of parenteral products. Pharm Dev Technol. 2009;14:567-587. 12. Patel SM, Bhugra C, Pikal MJ. Reduced pressure ice fog technique for controlled ice nucleation during freeze-drying. AAPS PharmSciTech. 2009;10:1406-1411. 13. Searles JA, Carpenter JF, Randolph TW. The ice nucleation temperature determines the primary drying rate of lyophilization for samples frozen on a temperature-controlled shelf. J Pharm Sci. 2001;90:860-871. 14. Patel SM, Doen T, Pikal MJ. Determination of end point of primary drying in freeze-drying process control. AAPS PharmSciTech. 2010;11:73-84. 15. Nail SL, Johnson W. Methodology for in-process determination of residual water in freeze-dried products. Dev Biol Stand. 1992;74:137-150 [discussion: 150-151].
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433 16. Tang XC, Nail SL, Pikal MJ. Evaluation of manometric temperature measurement (MTM), a process analytical technology tool in freeze drying, part III: heat and mass transfer measurement. AAPS PharmSciTech. 2006;7:97. 17. Tang X, Nail SL, Pikal MJ. Evaluation of manometric temperature measurement, a process analytical technology tool for freeze-drying: part I, product temperature measurement. AAPS PharmSciTech. 2006;7:E14. 18. Tang XC, Nail SL, Pikal MJ. Evaluation of manometric temperature measurement, a process analytical technology tool for freeze-drying: part II measurement of dry-layer resistance. AAPS PharmSciTech. 2006;7:93. 19. Tang XC, Nail SL, Pikal MJ. Freeze-drying process design by manometric temperature measurement: design of a smart freeze-dryer. Pharm Res. 2005;22: 685-700. 20. Hatley RH. The effective use of differential scanning calorimetry in the optimisation of freeze-drying processes and formulations. Dev Biol Stand. 1992;74: 105-119 [discussion: 119-122]. 21. Schersch K, Betz O, Garidel P, Muehlau S, Bassarab S, Winter G. Systematic investigation of the effect of lyophilizate collapse on pharmaceutically relevant proteins, part 2: stability during storage at elevated temperatures. J Pharm Sci. 2012;101:2288-2306. 22. Ullrich S, Seyferth S, Lee G. Measurement of shrinkage and cracking in lyophilized amorphous cakes. Part IV: effects of freezing protocol. Int J Pharm. 2015;495:52-57.
1433
€ rber C, Nunner B, Heschel I. The influence of the freezing process 23. Kochs M, Ko on vapour transport during sublimation in vacuum-freeze-drying. Int J Heat Mass Transfer. 1991;34:2395-2408. 24. Liu J, Viverette T, Virgin M, Anderson M, Paresh D. A study of the impact of freezing on the lyophilization of a concentrated formulation with a high fill depth. Pharm Dev Technol. 2005;10:261-272. 25. Nakagawa K, Hottot A, Vessot S, Andrieu J. Influence of controlled nucleation by ultrasounds on ice morphology of frozen formulations for pharmaceutical proteins freeze-drying. Chem Eng Process. 2006;45:783-791. 26. Kasper JC, Friess W. The freezing step in lyophilization: physico-chemical fundamentals, freezing methods and consequences on process performance and quality attributes of biopharmaceuticals. Eur J Pharm Biopharm. 2011;78: 248-263. 27. Liu J. Physical characterization of pharmaceutical formulations in frozen and freeze-dried solid states: techniques and applications in freeze-drying development. Pharm Dev Technol. 2006;11:3-28. 28. Patapoff TW, Overcashier DE. The importance of freezing on lyophilization cycle development. BioPharm Int. 2002;15(3):p16. 29. Izutsu K, Yonemochi E, Yomota C, Goda Y, Okuda H. Studying the morphology of lyophilized protein solids using X-ray micro-CT: effect of post-freeze annealing and controlled nucleation. AAPS PharmSciTech. 2014;15:1181-1188.
Contents lists available at ScienceDirect
Journal of Pharmaceutical Sciences journal homepage: www.jpharmsci.org
Pharmaceutical Biotechnology
Effect of Freezing on Lyophilization Process Performance and Drug Product Cake Appearance Reza Esfandiary, Shravan K. Gattu, John M. Stewart, Sajal M. Patel* Department of Formulation Sciences, Biopharmaceutical Development, MedImmune, Gaithersburg, Maryland 20878
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 November 2015 Revised 8 January 2016 Accepted 2 February 2016
This study highlights the significance of the freezing step and the critical role it can play in modulating process performance and product quality during freeze-drying. For the model protein formulation evaluated, the mechanism of freezing had a significant impact on cake appearance, a potential critical product quality attribute for a lyophilized drug product. Contrary to common knowledge, a freezing step with annealing resulted in 20% increase in primary drying time compared to without annealing. In addition, annealing resulted in poor cake appearance with shrinkage, cracks, and formation of a distinct skin at the top surface of the cake. Finally, higher product resistance (7.5 cm2.Torr.hr/g) was observed in the case of annealing compared to when annealing was not included (5 cm2.Torr.hr/g), which explains the longer primary drying time due to reduced sublimation rates. An alternative freezing option using controlled ice nucleation resulted in reduced primary drying time (i.e., 30% reduction compared to annealing) and a more homogenous batch with elegant uniform (i.e., significantly improved) cake appearance. Here, a mechanistic understanding of the distinct differences in cake appearance as a function of freezing mechanism is proposed within the context of ice nucleation temperature, ice crystal growth, and presumed solute distribution within the frozen matrix. © 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
Keywords: freeze-drying/lyophilization drying glass transition solid-state stability porosity processing process analytical technology (PAT) protein formulation
Introduction Freeze-drying is a commonly used unit operation to achieve acceptable shelf life for an otherwise unstable molecule. Freezedrying is usually carried out in three steps: freezing (i.e., conversion of water into ice), primary drying (i.e., removal of ice by sublimation at low temperature and pressure), and secondary drying (i.e., removal of unfrozen water). Typically, water content at the end of the secondary drying step is 1%. The ice nucleation temperature (or degree of supercooling) is a potential critical process parameter during the freezing step.1 The pore structure in the dried matrix is a reflectance of how the solution was frozen with every 1 C increase in ice nucleation temperature resulting in a 1%-3% reduction in primary drying time.2 A higher degree of supercooling results in smaller ice crystals, which results in higher product resistance and longer primary drying time but shorter secondary drying (due to higher specific surface area [SSA]). Thus, the freezing step affects not only primary drying but also secondary drying. The freezing step * Correspondence to: Sajal M. Patel (Telephone: þ1-301-398-5247; Fax: þ1-301398-7782). E-mail address: [email protected] (S.M. Patel).
of the freeze-drying process has received significant attention in the last decade because this first step of the process governs not only the performance of the subsequent steps (i.e., primary and secondary drying) but also the product quality (i.e., physical stability, cake appearance, residual moisture, reconstitution time).3 The ice nucleation temperature, random or stochastic in nature, is also an important process development and scale-up issue. Vials within a batch may vary in ice nucleation temperature from as high as 2 C to as low as 18 C. Thus, there is heterogeneity in ice nucleation temperature not only within a batch but also from batch to batch. An additional consideration is the clean room (Class 100) environment in manufacturing compared to laboratory bench preparation for a lab-scale dryer, resulting in differences in ice nucleation temperature.2 Annealing during freezing is often a common approach to mitigate heterogeneity in ice nucleation temperature, reduce primary drying time,3 and crystallize any crystallizing excipients in the formulation.4 During annealing, product temperature is maintained, for a certain period of time, above the glass transition temperature (Tg’) but below the melting temperature of the formulation matrix, resulting in Ostwald ripening (i.e., growth of bigger ice crystals at the expense of smaller ones). Although optimization of annealing time and temperature is critical, annealing
http://dx.doi.org/10.1016/j.xphs.2016.02.003 0022-3549/© 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
1428
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
may not always be suitable from a product quality standpoint due to conformational changes in protein structure or amorphousamorphous phase separation.5,6 In addition, for formulation matrices wherein crystallization of solute is involved, annealing may not always result in shorter primary drying time (i.e., lower dry layer resistance).7 Nonetheless, annealing during freezing is a way to mitigate the heterogeneity in ice nucleation temperature. Other alternative options have become available in the recent past to control ice nucleation temperature.8-11 The effect of ice nucleation temperature on process performance (i.e., primary drying time) using controlled ice nucleation techniques is well established.1,3,12,13 The SSA of the dried matrix is a good surrogate of the initial freezing characteristics of a solution. Higher ice nucleation temperature results in lower SSA and hence shorter primary drying time. Developing an understanding of the effect of controlled ice nucleation on product quality is now becoming critical to enable controlled ice nucleation techniques as process analytical tools to control ice nucleation temperature in clinical and commercial batches. This work, presented as an industry case study, assesses the negative effect of annealing on cake appearance and demonstrates how controlled ice nucleation results in significant improvement in cake appearance as well as batch homogeneity. Cake appearance is a potential critical product quality attribute with a specification both for lot release and stability. The finished product is visually inspected (100%) for cake appearance because a change or variability in cake appearance from what is anticipated is likely an indication of change in product quality. Although the acceptance criterion for cake appearance is outside the scope of this work, the study here clearly demonstrates the significance of the freezing step and its impact on cake appearance.
(SP Scientific, Stone Ridge, NY) at 20% of total lyophilizer shelf load. A 15-min equilibration at 5 C was used before initiation of the freeze-drying cycle. Although different freezing methods were used (Table 1), the primary drying (30 C, 100 mTorr) and secondary drying (40 C, 100 mTorr) steps were kept identical for all the cycles tested. The pressure was controlled using a capacitance manometer gauge. The duration of the secondary drying step was fixed at 6 h in all the freeze-drying cycles whereas the duration/end of the primary drying step was determined by a comparative pressure measurement wherein the Pirani gauge pressure measurement converges, within 10 mT, with that of the capacitance manometer chamber pressure.14,15 The product temperature during freeze-drying was monitored using thermocouples (TCs) and also determined via Manometric temperature measurement (MTM). A total of 4 TCs were placed at the bottom center of selected vials located at the edge (2 vials) and center (2 vials) of each shelf. The TCs were distributed across the shelf such that any variation in product temperature as a function of shelf location was monitored. To evaluate the impact of the freezing step on overall process efficiency and product quality, the freeze-drying cycles were designed to differ in the freezing step. One of the freeze-drying cycles included an annealing step with a temperature set point of 15 C (referred to as Anneal), whereas the other cycle used the ControLyo™ Technology10 for a controlled ice nucleation step (referred as CN) at a temperature set point of 8 C in place of annealing. Finally, a third cycle without either annealing or controlled ice nucleation was evaluated (referred to as No Anneal). The freeze-drying process parameters for all three cycles are summarized in Table 1. Determination of Product Resistance and Product Temperature Using MTM Measurement
Materials and Methods Materials The model protein (referred as Protein X) was produced and purified at MedImmune (Gaithersburg, MD). Protein X is a glycoprotein with a size of ~75 kDa, formulated at 10 mg/mL in a citrate-based buffer containing trehalose dihydrate, an arginine/arginine-HCl mixture, and polysorbate 80 (PS80) as excipients. All excipients used were of compendial pharmaceutical grade. Lyophilization Process Parameters For all freeze-drying experiments, 3.5 mL of 0.22-mm filtered Protein X formulation was filled aseptically into 10-mL vials and partially stoppered with 20-mm stoppers (both the vials and stoppers were autoclaved and sufficiently dried before use). The freeze-drying cycles were performed on a Lyostar 3 freeze-dryer
MTM involved closing of the valve connecting the chamber and the condenser for a brief period (~25 s) and recording the pressure rise as a function of time. The pressure rise data were fitted to the MTM equation to determine product temperature at the sublimation interface, product resistance, and vial heat transfer coefficient.16-19 Differential Scanning Calorimetry Differential Scanning Calorimeter Q2000 series from TA instruments (New Castle, DE) was used to determine the glass transition temperature (Tg’). For the Tg’ measurement, 20 mL of liquid sample was added into an aluminum pan and sealed hermetically. An empty pan and lid were used as the reference pan. The sample was frozen to 60 C and then heated to 25 C at a rate of 5 C/min. The Tg’ value was determined using Universal Analysis software (TA instruments, New Castle, DE) and reported as the midpoint of the glass transition.20
Table 1 Lyophilization Process Parameters for the Anneal, No Anneal, and Controlled Ice Nucleation (CN) Cycle Process Parameter/Cycle
Anneal Cycle
No Anneal Cycle
CN Cycle
Ramp to freezing Freezing temperature/hold time Annealing temperature/hold time Controlled ice nucleation temperature/hold time Ramp to primary drying Primary drying temperature/pressure/hold time Ramp to secondary drying Secondary drying temperature/pressure/hold time
1 C/min 40 C/4 h 15 C/2 h NA 0.1 C/min 30 C/100 mTorr/Pirani 0.3 C/min 40 C/100 mTorr/6 h
1 C/min 40 C/4 h NA NA 0.1 C/min 30 C/100 mTorr/Pirani 0.3 C/min 40 C/100 mTorr/6 h
1 C/min 40 C/4 h NA 8 C/4 hr 0.1 C/min 30 C/100 mTorr/Pirani 0.3 C/min 40 C/100 mTorr/6 h
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
1429
Residual Moisture Determination A Mettler Toledo C30 coulometric KF titrator equipped with a Stromboli oven sample changer (Columbus, OH) was used for residual moisture measurements. In brief, the lyophilized cake was broken down using a spatula. About 100 mg of sample was transferred to a clean 20-mL vial and sealed. A blank (i.e., an empty vial) was also prepared and used to equilibrate the system until the drift was below the intended 25-mg water/min baseline value. System suitability was assessed using a reference standard with target 1% residual moisture content. All sample preparations were performed in the glove box with nitrogen purge (< 0.5% relative humidity). Samples were heated for 10 min at 100 C and residual moisture was calculated using the Lab-X software (Mettler Toledo). Scanning Electron Microscopy Scanning electron microscopy (SEM) was used for qualitative assessment of the lyophilized cake morphology, more specifically the pore structure. The lyophilized cake was first cut into large pieces using a clean sharp blade. These pieces were then sliced along the cylindrical axis. The untouched cake surface was then used for SEM examination by a bench top SEM instrument (Model TM-1000, Hitachi, Clarksburg, MD). All samples were sputter coated using gold with 10-nm thickness using a sputter coater from Electron Microscopy Sciences (Model EMS 150R ES, Hatfield, PA). Results Thermal Analysis The glass transition temperature (Tg’) of the maximally freezeconcentrated Protein X formulation was determined using differential scanning calorimetry as detailed in the Methods section. The Tg’ was determined to be about 34 C without any crystallization event (data not shown). Lyophilization Process Data The Protein X formulation was lyophilized using three uniquely designed freeze-drying cycles (Anneal, No Anneal, and CN) with distinct differences in the freezing step. These cycles, as detailed in Table 1, were identical in the primary and secondary drying steps and were run at the same batch size. The impact of the freezing step/mechanism is discussed in the following section with regard to both process performance as well as product quality.
Figure 1. Product temperature determined by MTM during primary drying for the Anneal, No Anneal, and CN cycles.
Primary Drying Time The end of primary drying was defined when the Pirani pressure converged within 10 mTorr of the capacitance manometer. The Pirani pressure profile for the Anneal, No Anneal, and CN cycles is shown in Figure 2. The Anneal cycle resulted in the longest primary drying (about 120 h). The primary drying time was reduced by about 20% and 30% in the No Anneal (about 95 h) and CN cycles (about 85 h), respectively. The significant differences observed in primary drying time are exclusively due to freezing differences because all other parameters (formulation, batch size, vial and stopper type, fill volume, and primary and secondary drying parameters) were kept the same across all the cycles tested. Another interesting observation is the difference in slopes of the Pirani pressure drop with values of about 4, 2.5, and 1.5 mTorr/hr for the Anneal, No Anneal, and CN cycles, respectively. The steepness of the slope of the Pirani pressure is an indication of batch homogeneity, where a steeper slope indicates a more homogeneous batch. Thus, the batch homogeneity follows the trend CN > No Anneal > Anneal.
Process Performance Product Temperature. Product temperature during each freezedrying cycle was monitored using TCs and determined using MTM methodology. The product temperature during primary drying as determined by MTM is compared in Figure 1. During the steady-state stage of the primary drying, the product temperature was similar for No Anneal and CN cycles at about 35.5 C, whereas for the Anneal cycle the product temperature was about 1 C warmer (about 34.5 C). The general trend in product temperature during primary drying followed Anneal > No Anneal > CN. Under all conditions, product temperature was close to or below Tg’ (the collapse temperature, measured using freeze dry microscopy, was within 0.5 C of Tg’, data not shown).
Figure 2. The Pirani chamber pressure during primary drying for the Anneal, No Anneal, and CN cycles. The spikes in the pressure are due to MTM measurement.
1430
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
Product Quality Cake Appearance. Figure 3 shows representative pictures of the lyophilized cakes obtained from the Anneal, No Anneal, and CN cycles with distinct differences observed as a function of the freezing conditions tested. The lyophilized cakes in the Anneal cycle exhibited the poorest quality with extensive cake shrinkage at the bottom of the vials, cracks, and collapse, as well as an enhanced level of porousness and a distinct skin formation at the top of the dried cakes. Samples from the No Anneal cycle, although relatively improved in their appearance compared to the Anneal cycle, still exhibited partial collapse at the bottom of the vial but with no visible skin formation at the top. The vials lyophilized using the CN cycle resulted in elegant cakes with no cake defects. In general, the described cake appearance characteristics within each cycle were consistent across all the vials irrespective of the vial location on the shelf. Product Resistance Product (or dry layer) resistance, a reflection of how the solution is frozen, governs the drying rate (i.e., mass transfer), which in turn governs the product temperature (i.e., heat transfer). To better understand the potential impact of the different freezing methodologies on product resistance, MTM was used to determine product resistance as a function of dry layer thickness. Data showed that vials subjected to annealing (Anneal) exhibited the highest level of product resistance at about 7.5 cm2 Torr h/g while lower values of about 5 and 4 cm2 Torr h/g (product resistance reported at about 0.5 cm of dry layer thickness) were observed for the No Anneal and CN cycles, respectively (Fig. 4). The observed product resistance values correlated well with the primary drying time in Figure 2, with a progressive reduction in product resistance directly correlating with shorter primary drying time. The Anneal cycle with the highest dry layer resistance showed the longest primary drying time of about 120 h, whereas the No Anneal and CN cycles with lower resistance values resulted in shorter primary drying durations of about 95 h and 85 h, respectively. In addition, a unique characteristic of the product resistance data specific to the Anneal
cycle was the significantly high product resistance at the start of drying followed by a decrease in product resistance with increasing dry layer thickness, a characteristic of skin formation followed by micro-collapse/collapse, as supported by the cake appearance (Fig. 3). Scanning Electron Microscopy A qualitative examination of the lyophilized cakes by SEM demonstrated significant differences in pore structure when comparing samples from the Anneal cycle to those from No Anneal and CN cycles. Although the samples, imaged from top cross-sections of the lyophilized cakes in the No Anneal and CN cycles, exhibited similar and perhaps more typical pore structures, those obtained from the Anneal cycle showed a distinct layer at the top, evident by the exhibited solid sheet (i.e., closed pore structure) only observed in case of annealing (Fig. 5). This observation is aligned with cake appearance observations (Fig. 3) where a distinct skin formation was noted for samples from the Anneal cycle. Residual Moisture and Stability of the Lyophilized Cakes The residual moisture content was determined to be 1% across all three cycles. To assess the impact of differences in cake appearance, product from Anneal, No Anneal, and CN cycles was tested for stability at 40 C for 1 month. There was no change in product quality prelyophilization and postlyophilization as well as on storage at 40 C up to 1 month (data not shown), in agreement with previously published work21 that collapse does not affect product stability. Discussion This work clearly highlights the importance of the freezing step in lyophilization of a model protein formulation, impacting both process performance and the product quality. The complex role of ice nucleation and ice crystal growth during freezing is evident from different freezing steps/mechanisms (i.e., annealing, no annealing, and controlled ice nucleation).3
Figure 3. Representative product cake appearance for the Annealed (left column), No Anneal (middle column), and Controlled Nucleation (right column) cycles. The pictures on the top row are captured from the vials positioned inverted whereas the ones on the bottom row are captured from the same vials positioned upright.
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
Figure 4. Product resistance as a function of dry layer thickness determined by MTM for the Anneal, No Anneal, and CN cycles.
Contrary to common knowledge regarding enhanced batch homogeneity and reduced primary drying time with use of annealing, for the particular formulation matrix studied in this work, annealing at 15 C for 120 min resulted in batch heterogeneity and extended primary drying time.7 Data here showed that use of annealing (Anneal) extended the primary drying duration by about 20% compared to No Anneal (Fig. 2). The longer primary drying time in case of annealing can be explained based on the observed higher product resistance (Fig. 4) and skin formation (Figs. 3 and 5) causing reduced sublimation rate. In addition, annealing resulted in skin formation followed by cracks and collapse,22 during primary drying as suggested by the gradual decrease in product resistance values (Fig. 4). Use of controlled ice nucleation proved to be an effective process optimization approach where about 30% reduction in primary drying time was observed compared to the annealed cycle (Fig. 2). In addition, although exclusion of annealing resulted in some improvements in cake appearance with minor cake shrinkage and partial collapse still present, use of the controlled ice nucleation significantly enhanced cake appearance, resulting in elegant lyophilized cakes (Fig. 3). In the case of uncontrolled ice nucleation, the temperature is coldest (i.e., highest degree of supercooling) and the probability of ice nucleation and particulate matter (if any) is highest at the bottom of the vial, in contact with the temperature-controlled
1431
shelves. On ice nucleation, ice crystals grow very rapidly into the supercooled solution. The product temperature then increases (typically up to the equilibrium freezing temperature of the solution) as the heat of crystallization is released from the frozen sample to the cold bottom of the vial. After this initial ice nucleation, the ice crystals grow rapidly toward the surface of the sample resulting in columnar, dendritic, or finger-like ice crystal morphology.23 During annealing, as large ice crystals are formed at the expense of the smaller ice crystals (i.e., Oswald ripening), the highly concentrated solutes trapped between the ice crystals are expelled via a convective flow toward the ice finger tips. Usually the solutes are distributed homogeneously during ice nucleation (Fig. 6, No Anneal cycle); however, they macroscopically redistribute during ice crystal growth from regions of higher concentration toward larger forming ice crystals via convective flow. The ice growth is halted when it reaches the surface of the sample, further resulting in the flow of highly concentrated solutes out of the interstices. This produces a thick layer of concentrated solute on the surface of sample, exhibiting a distinct skin formation after freeze-drying (Figs. 3 and 5, Anneal cycle). This hypothesis is supported by the observation of skin formation on annealing, whereas no skin formation was observed in the absence of the annealing step. The minor collapse observed in the No Anneal cycle could be due to product temperature being close to or exceeding Tg'. In the case of controlled ice nucleation,10 ice first nucleates at the top of the sample followed by a downward propagation, resulting in uniform ice crystal growth and homogeneous distribution of the solutes within the interstices. This is clearly supported by the lower product resistance (i.e., higher porosity) in Figure 4, lower product temperature (Fig. 1), shorter primary drying time (Fig. 2), and elegant cake appearance (Fig. 3). On the contrary, if skin formation were to occur with controlled ice nucleation (as a worstcase scenario), it would be formed at the bottom of the vial (i.e., last to dry), which very likely would go unnoticed during visual inspection because it would have minimal to no impact on the macroscopic cake appearance. This work clearly highlights the significance of the freezing step and the critical role it can play in modulating the process performance24-26 as well as product quality during lyophilization. As detailed in the previous section, the mechanism of freezing has a significant impact on cake appearance, a potential critical product quality attribute for lyophilized products. For the particular case study described here, no impact on stability (i.e., other product quality attributes) was observed as a function of cake appearance (i.e., Anneal, No Anneal, and CN), which is in line with prior publications wherein even a collapsed cake had comparable stability to a noncollapsed cake.21 One can, however, question if mitigating cake appearance defects is critical at the expense of additional resources particularly when stability is not of concern. If one were to accept cake defects, then demonstrating intrabatch and interbatch
Figure 5. Representative scanning electron microscopy (SEM) images of lyophilized cakes from the Annealed (left), No Anneal (middle), and Controlled Nucleation (right) cycles. All images are obtained from the top cross-section of the cakes.
1432
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433
Figure 6. Representative schematics of the ice crystal structures for the No Anneal (left), Control Nucleation (middle), and Anneal (right) cycles.
reproducibility still remains critical. Clearly, differences in the freezing step here resulted in significantly different cake appearance, which is a concern because freezing differences do exist between laboratory and production scales, at least in the case of uncontrolled ice nucleation. A clear distinction must be made between poor process control/understanding versus formulation and process limitations to decide whether a cake defect is acceptable. Nonetheless, as mentioned earlier, defining and justifying acceptance criteria for cake appearance are not the focus of this work. In general, to arrive at more widely applied conclusions regarding a potential correlation between cake appearance and physical stability or lack thereof, more comprehensive assessment is needed over a wider formulation and process design space as well as evaluation of real-time, long-term stability under intended storage conditions. The need for a solid-state characterization technique that can help evaluate the subtle differences in cake morphology/pore structure at microscopic level is critical to assess the effect of freezing parameters on finished product quality attributes, specifically cake appearance.27 Although SEM provides visualization of the cake morphology, it is still qualitative and depends on sample preparation. SSA provides a quantitative measurement as measured via the Brunauer-Emmett-Teller technique, but Brunauer-Emmett-Teller is very unlikely to pick up small differences in SSA due to subtle freezing differences. The parafin dye method28 allows visualization of the cake morphology but requires beforehand sample preparation and cake sectioning to view under the microscope. The X-ray micro-CT29 is an emerging technology to evaluate morphology of freeze-dried cakes. A nondestructive technique with the capability to perform quantitative analysis and that can provide visualization of the inside of the cake would be immensely useful to characterize the cake morphology after lyophilization. Another area that needs additional research is the impact of formulation matrices on the freeze-drying process (i.e., why skin formation is observed on annealing with some formulations and not with others). Annealing has been applied during freezing for several formulation and process conditions, and usually no skin formation is observed. However, the presence of protein and the complex interactions within different formulation matrices on freezing needs to be evaluated to better understand the effect of formulation matrix on freezing mechanism. Additionally, the effect of fill volume and vial size (i.e., drug product presentation) needs to be evaluated to identify critical process parameters that govern freezing mechanism. Identifying physical properties of the solution that govern freezing mechanism and establishing correlation with product quality will help build tools that would enable predicting the effect of process parameters on product quality. In future studies, we intend on evaluating formulation and fill-finish parameters using the same model protein to establish the link between formulation and process and its impact on drug product quality.
Conclusion This work clearly demonstrates the effect of the freezing step on lyophilized drug product cake appearance. Annealing, commonly applied during the freezing step, resulted in skin formation, for the conditions evaluated, causing product collapse (at the extreme extent). The design of the freezing step (annealing, no annealing, or controlled ice nucleation) indirectly governs the freezing mechanism, which may further impact product quality (in this case cake appearance). Although primary drying continues to receive the highest attention from process optimization and scale-up perspectives, the criticality of the freezing step in the freeze-drying process design, development, optimization, and scale-up is not to be ignored. Acknowledgments The authors thank Nancy Craighead for the critical review of this manuscript. References 1. Rambhatla S, Ramot R, Bhugra C, Pikal MJ. Heat and mass transfer scale-up issues during freeze drying: II. Control and characterization of the degree of supercooling. AAPS PharmSciTech. 2004;5:e58. 2. Roy ML, Pikal MJ. Process control in freeze drying: determination of the end point of sublimation drying by an electronic moisture sensor. J Parenter Sci Technol. 1989;43:60-66. 3. Searles JA, Carpenter JF, Randolph TW. Annealing to optimize the primary drying rate, reduce freezing-induced drying rate heterogeneity, and determine T(g)' in pharmaceutical lyophilization. J Pharm Sci. 2001;90:872-887. 4. Li X, Nail SL. Kinetics of glycine crystallization during freezing of sucrose/ glycine excipient systems. J Pharm Sci. 2005;94:625-631. 5. Heller MC, Carpenter JF, Randolph TW. Manipulation of lyophilization-induced phase separation: implications for pharmaceutical proteins. Biotechnol Prog. 1997;13:590-596. 6. Lueckel B, Helk B, Bodmer D, Leuenberger H. Effects of formulation and process variables on the aggregation of freeze-dried interleukin-6 (IL-6) after lyophilization and on storage. Pharm Dev Technol. 1998;3:337-346. 7. Lu X, Pikal MJ. Freeze-drying of mannitol-trehalose-sodium chloride-based formulations: the impact of annealing on dry layer resistance to mass transfer and cake structure. Pharm Dev Technol. 2004;9:85-95. 8. Geidobler R, Winter G. Controlled ice nucleation in the field of freeze-drying: fundamentals and technology review. Eur J Pharm Biopharm. 2013;85:214-222. 9. Kasper JC, Winter G, Friess W. Recent advances and further challenges in lyophilization. Eur J Pharm Biopharm. 2013;85:162-169. 10. Konstantinidis AK, Kuu W, Otten L, Nail SL, Sever RR. Controlled nucleation in freeze-drying: effects on pore size in the dried product layer, mass transfer resistance, and primary drying rate. J Pharm Sci. 2011;100:3453-3470. 11. Patel SM, Pikal M. Process analytical technologies (PAT) in freeze-drying of parenteral products. Pharm Dev Technol. 2009;14:567-587. 12. Patel SM, Bhugra C, Pikal MJ. Reduced pressure ice fog technique for controlled ice nucleation during freeze-drying. AAPS PharmSciTech. 2009;10:1406-1411. 13. Searles JA, Carpenter JF, Randolph TW. The ice nucleation temperature determines the primary drying rate of lyophilization for samples frozen on a temperature-controlled shelf. J Pharm Sci. 2001;90:860-871. 14. Patel SM, Doen T, Pikal MJ. Determination of end point of primary drying in freeze-drying process control. AAPS PharmSciTech. 2010;11:73-84. 15. Nail SL, Johnson W. Methodology for in-process determination of residual water in freeze-dried products. Dev Biol Stand. 1992;74:137-150 [discussion: 150-151].
R. Esfandiary et al. / Journal of Pharmaceutical Sciences 105 (2016) 1427e1433 16. Tang XC, Nail SL, Pikal MJ. Evaluation of manometric temperature measurement (MTM), a process analytical technology tool in freeze drying, part III: heat and mass transfer measurement. AAPS PharmSciTech. 2006;7:97. 17. Tang X, Nail SL, Pikal MJ. Evaluation of manometric temperature measurement, a process analytical technology tool for freeze-drying: part I, product temperature measurement. AAPS PharmSciTech. 2006;7:E14. 18. Tang XC, Nail SL, Pikal MJ. Evaluation of manometric temperature measurement, a process analytical technology tool for freeze-drying: part II measurement of dry-layer resistance. AAPS PharmSciTech. 2006;7:93. 19. Tang XC, Nail SL, Pikal MJ. Freeze-drying process design by manometric temperature measurement: design of a smart freeze-dryer. Pharm Res. 2005;22: 685-700. 20. Hatley RH. The effective use of differential scanning calorimetry in the optimisation of freeze-drying processes and formulations. Dev Biol Stand. 1992;74: 105-119 [discussion: 119-122]. 21. Schersch K, Betz O, Garidel P, Muehlau S, Bassarab S, Winter G. Systematic investigation of the effect of lyophilizate collapse on pharmaceutically relevant proteins, part 2: stability during storage at elevated temperatures. J Pharm Sci. 2012;101:2288-2306. 22. Ullrich S, Seyferth S, Lee G. Measurement of shrinkage and cracking in lyophilized amorphous cakes. Part IV: effects of freezing protocol. Int J Pharm. 2015;495:52-57.
1433
€ rber C, Nunner B, Heschel I. The influence of the freezing process 23. Kochs M, Ko on vapour transport during sublimation in vacuum-freeze-drying. Int J Heat Mass Transfer. 1991;34:2395-2408. 24. Liu J, Viverette T, Virgin M, Anderson M, Paresh D. A study of the impact of freezing on the lyophilization of a concentrated formulation with a high fill depth. Pharm Dev Technol. 2005;10:261-272. 25. Nakagawa K, Hottot A, Vessot S, Andrieu J. Influence of controlled nucleation by ultrasounds on ice morphology of frozen formulations for pharmaceutical proteins freeze-drying. Chem Eng Process. 2006;45:783-791. 26. Kasper JC, Friess W. The freezing step in lyophilization: physico-chemical fundamentals, freezing methods and consequences on process performance and quality attributes of biopharmaceuticals. Eur J Pharm Biopharm. 2011;78: 248-263. 27. Liu J. Physical characterization of pharmaceutical formulations in frozen and freeze-dried solid states: techniques and applications in freeze-drying development. Pharm Dev Technol. 2006;11:3-28. 28. Patapoff TW, Overcashier DE. The importance of freezing on lyophilization cycle development. BioPharm Int. 2002;15(3):p16. 29. Izutsu K, Yonemochi E, Yomota C, Goda Y, Okuda H. Studying the morphology of lyophilized protein solids using X-ray micro-CT: effect of post-freeze annealing and controlled nucleation. AAPS PharmSciTech. 2014;15:1181-1188.