Journal of Pharmaceutical Sciences 105 (2016) 40e49
Contents lists available at ScienceDirect
Journal of Pharmaceutical Sciences journal homepage: www.jpharmsci.org
Pharmaceutical Biotechnology
Freeze-Drying Above the Glass Transition Temperature in Amorphous Protein Formulations While Maintaining Product Quality and Improving Process Efficiency Roberto A. Depaz, Swapnil Pansare, Sajal Manubhai Patel* Department of Formulation Sciences, Biopharmaceutical Development, MedImmune, 1 MedImmune Way, 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 3 June 2015 Revised 31 August 2015 Accepted 5 October 2015 Available online 18 November 2015
This study explored the ability to conduct primary drying during lyophilization at product temperatures above the glass transition temperature of the maximally freeze-concentrated solution (Tg0 ) in amorphous formulations for four proteins from three different classes. Drying above Tg0 resulted in significant reductions in lyophilization cycle time. At higher protein concentrations, formulations freeze dried above Tg0 but below the collapse temperature yielded pharmaceutically acceptable cakes. However, using an immunoglobulin G type 4 monoclonal antibody as an example, we found that as protein concentration decreased, minor extents of collapse were observed in formulations dried at higher temperatures. No other impacts to product quality, physical stability, or chemical stability were observed in this study among the different drying conditions for the different proteins. Drying amorphous formulations above Tg0 , particularly high protein concentration formulations, is a viable means to achieve significant time and cost savings in freeze-drying processes. © 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
Keywords: freeze-drying lyophilization protein formulation thermal analysis glass transition stability processing drying
Introduction Freeze-drying, or lyophilization, is a unit operation commonly used to stabilize protein formulations to achieve longer shelf life, compared with solution formulations, during product development and commercial distribution. In the last decade, more than 40% of marketed biotherapeutics were freeze dried,1 and with new complex molecular formats being developed (e.g., bispecific molecules, fusion proteins, antibody drug conjugates), this percentage is expected to further increase. However, the processes of freezing and drying could each be detrimental to protein stability. Thus, optimization of the formulation composition so that the protein is stable against the stresses imposed by freeze-drying is a key element of drug product development.2-4 Additionally, optimization of the freeze-drying process is equally critical to minimize
Abbreviations used: HPSEC, high-performance size-exclusion chromatography; IgG1, immunoglobulin G type 1; IgG4, immunoglobulin G type 4; mAb, monoclonal antibody; MFI, microflow imaging; SbVP, subvisible particle; Tc, collapse temperature determined by freeze-dry microscopy; Tg0 , glass transition temperature of the maximally freeze-concentrated solution; Ts, shelf temperature. * Correspondence to: Sajal M. Patel (Telephone: 301-398-5247; Fax: 301-3987782). E-mail address: [email protected] (S.M. Patel).
processing time for increased throughput and also to reduce production costs.5 A typical freeze-drying process consists of three steps: (1) freezing wherein water is converted into ice; (2) primary drying wherein ice is removed by sublimation; and (3) secondary drying wherein unfrozen water is removed by desorption. Typically, at the end of secondary drying the residual water content is less than 1%. Of these three steps, primary drying is often the longest step and hence optimization of this step is usually the main focus in industry. During primary drying, the product temperature needs to be maintained 2 Ce3 C below the maximum allowable temperature, which for an amorphous matrix is the glass transition temperature of maximally freeze-concentrated solution (Tg0 ) or the collapse temperature (Tc).6 Drying above Tg0 or Tc may result in cake collapse that could further impact product quality attributes such as residual water content, reconstitution time, and protein stability.7-9 Usually, Tc is within 1 Ce2 C of Tg0 , and Tg0 and Tc may be used interchangeably.10 However, for higher concentration protein formulations, visible collapse may not always be observed when drying above Tg0 as the difference between Tg0 and the Tc may increase as protein concentration increases.11 Recent publications have shown that even with total collapse, protein stability either improves or is not significantly different compared with that in the uncollapsed cake.12-14 However, cake
http://dx.doi.org/10.1002/jps.24705 0022-3549/© 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
41
and stoppers were obtained from West Pharmaceutical Services, Inc. (Exton, Pennsylvania).
appearance is a potential critical quality attribute for a lyophilized product and hence included in product specifications for lot release and stability. In general, improvement in process efficiency at the expense of product appearance may not be desired. Thus, although drying above the Tc may not be practical from an appearance perspective, drying above Tg0 is a very practical approach to reduce freeze-drying cycle time without impacting product quality,11 especially for high concentration protein formulations. Furthermore, for existing processes where primary drying occurs at product temperatures below Tg0 , data supporting maintained product quality if product temperature exceed Tg0 during primary drying is invaluable in the event of any temperature and pressure excursions that may occur during production. On the basis of the Fox equation and the relatively high Tg0 values of proteins, the Tg0 of a protein-containing amorphous formulation is expected to increase with protein concentration assuming an otherwise constant formulation composition.15,16 However, understanding the effect that increasing protein concentration may have on Tc, and the extent to which Tc may differ from Tg0 , is critical in defining the maximum allowable product temperature during primary drying. Because every 1 C increase in product temperature during primary drying may result in approximately 13% reduction in primary drying time,6 operating at the highest allowable product temperature is highly desired. Thus, thermal characterization of the formulation by differential scanning calorimetry (DSC) for Tg0 and freeze-dry microscopy (FDM) for Tc is critical for both product quality and process efficiency. In this work, both product quality and process performance were evaluated when conducting primary drying at product temperatures above and below Tg0 for four different proteins [two immunoglobulin G type 1 (IgG1) monoclonal antibodies, an immunoglobulin G type 4 (IgG4) monoclonal antibody (mAb), and a fusion protein] in an amorphous formulation matrix. Different primary drying conditions were systematically evaluated, and stability studies were performed to show there was no impact to product quality both upon lyophilization and also during long-term storage. The relationship between Tg0 and Tc as a function of protein concentration is presented for the various molecules, indicating the potential to exceed product temperature above Tg0 during primary drying, particularly at higher protein concentrations, to achieve significant reduction in lyophilization cycle time.
Vials were partially stoppered with 13 mm (for 3-mL vials) or 20 mm (for 5-mL and 20-mL vials) lyophilization stoppers. Freezedrying was performed using a Virtis Genesis 35 EL freeze dryer (SP Scientific, Stone Ridge, New York) or Millrock PDQ24XS-S freeze dryer (Millrock Technology, Kingston, New York). Vials were loaded onto the freeze dryer at a shelf temperature (Ts) of 20 C. After loading the shelves, they were cooled to 5 C and held for 30 min. The shelves were then cooled to 5 C and held for 15 min before cooling to 40 C. All molecules except mAb C underwent an annealing step at 16 C for 120 min before bringing back to 40 C. The annealing time and temperature was selected based on process optimization activities leading to batch homogeneity in terms of drying and product quality. After holding at 40 for 2 h, the Ts was raised to the desired primary drying Ts (see Results section) where primary drying occurred at a chamber pressure of 100 mTorr. Chamber pressure was controlled using the Capacitance Manometer gauge. End of primary drying was determined by comparative pressure measurement wherein the Pirani gauge pressure measurement converges with that of the Capacitance Manometer chamber pressure.17 Secondary drying was performed at 40 C for 6 h while maintaining the pressure at 100 mTorr. Product temperature was monitored during the freeze-drying process using thermocouples placed at the bottom center of selected vials. The variability in product temperature as measured by the thermocouples was ±0.5 C.18 With the following two exceptions, all Ts ramps throughout the cycle were conducted at 0.5 C/min: (1) for mAb A, the ramp rate from 40 C to the primary drying Ts was 0.1 C/min, and (2) all ramp rates during secondary drying to 40 C were 0.1 C/min.
Materials and Methods
Differential Scanning Calorimetry
Materials
A Q2000 series differential scanning calorimeter from TA Instruments (New Castle, Delaware) was used for glass transition temperature measurements. For Tg0 measurements, 20 mL of liquid sample was added into an aluminum pan and sealed hermetically. An empty pan and lid was used as the reference. Liquid samples were frozen to 60 C at a rate of 5 C/min and then heated to 25 C at a rate of 5 C/min. The Tg0 values were determined using Universal Analysis software and reported as the midpoint of the glass transition.19-21 For Tg determination of the lyophilized cakes, approximately 5 mg of sample were sealed into an aluminum pan and analyzed in modulated DSC mode. Pans were heated at 5 C/min through the glass transition with a modulation period of 80 s and amplitude ±0.5 C.
The proteins evaluated in this work were produced and purified at MedImmune (Gaithersburg, Maryland) using proprietary methods. mAb A and mAb B are IgG1 monoclonal antibodies, and mAb C is an IgG4 mAb. Protein X (or “Pro X”) is a fusion protein with a molecular weight of approximately 90 kDa. Each of these molecules was formulated in a buffered solution with a disaccharide and polysorbate 80 as excipients. mAb B contained an amino acid as an additional excipient. Formulations contained excipients expected to remain amorphous upon lyophilization, and no components expected to crystallize were included. All excipients used were multicompendial grade. For thermal analysis, mAb A and mAb C samples were prepared in their formulation buffer at concentrations ranging between 1 and 100 mg/mL. For lyophilization, mAb A was prepared at 100 mg/mL, and mAb C was prepared at three different concentrations: 5, 25, and 100 mg/mL with the levels of all other formulation components remaining the same. mAb B and Pro X were prepared at 50 mg/mL for thermal analysis and lyophilization. Type 1 glass tubing vials (3-, 5-, and 20-mL), and chlorobutyl, single vent lyophilization stoppers were used. Both vials
Lyophilization For their individual lyophilization cycles, the proteins were filled into vials at the following conditions: mAb A and mAb C: 3-mL vial, 1.1 mL fill mAb B: 5-mL vial, 2.7-mL fill Pro X: 20-mL vial, 5.5-mL fill
Freeze-Dry Microscopy All liquid samples were tested using an Olympus BX50 microscope with a Linkam FDCS 196 stage. Images were recorded using a QImaging camera attachment. A sample volume of 5-10 mL was placed between two cover slips on the FDM stage. The samples were cooled and frozen to 40 C. All samples, with the exception of
42
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
mAb C, then underwent an annealing step at 16 C followed by refreeze to 40 C. A rate of 10 C/min was used throughout the freezing steps. After freezing, a vacuum at 100 mTorr was initiated using a vacuum pump, and heating was initiated at a rate of 1 C/min. The ramp rate was decreased to 0.5 C/min when the temperature exceeded Tg0 and was near the Tc. Onset of collapse was recorded as the Tc, when a change in the structure of the freezedried solid was observed as described elsewhere.22 High-Performance Size-Exclusion Chromatography High-performance size-exclusion chromatography (HPSEC) was performed using an Agilent HPLC system with UV absorbance at 280 nm. A 7.8 300 mm2 TSKgel G3000SWXL column with 6.0 40 mm2 TSKgel SWXL guard column (Tosoh Bio-science, King of Prussia, Pennsylvania) was used with a mobile phase comprising 0.1 M sodium phosphate, 0.1 M sodium sulfate, 0.05% (w/v) sodium azide, pH 6.8 (the mobile phase for Pro X was 90% of this solution along with 10% isopropanol). Formulations at 25, 50, and 100 mg/mL were diluted to 10 mg/mL with phosphate-buffered saline (Gibco Life Technologies Corporation, Grand Island, New York) prior to analysis, and 25 mL were injected. Formulations at 5 mg/mL were injected neat with an injection volume of 50 mL. The flow rate was 1 mL/min, and samples were held at 5 C in the autosampler tray prior to injection. The relative percentage of each species in each sample (aggregates, monomer, fragments) was calculated relative to the total area of all peaks. Moisture Determination Residual moisture content was determined using a Mettler Toledo DL39 Karl Fischer (KF) coulometer. Lyophilized product was reconstituted with anhydrous methanol and then injected into the KF coulometer for analysis. The percentage residual moisture content was determined based upon the amount of water titrated relative to the mass of the lyophilized sample. Microflow Imaging Subvisible particle (SbVP) analysis by microflow imaging (MFI) was performed using an MFI 5200 from Protein Simple (Santa Clara, California). Samples were analyzed neat. Prior to each sample measurement, the background illumination was optimized using formulation buffer. Approximately 0.25 mL sample was allowed to flow through prior to counting to generate a background free of Schlieren effects. Counts for particle sizes 2 mm, 10 mm, and 25 mm were recorded for particles with an aspect ratio less than 0.85.
Pro X, the bioassay measured the ability of the protein to inhibit binding of the respective molecule target to cells expressing a luciferase reporter gene. Luciferase expression was quantified via the use of a chemiluminescent substrate and an Envision plate reader (Perkin Elmer, Waltham, MA). For mAb B, the bioassay measured the ability of the protein to protect cells from cell death. The amount of luminescence measured after reaction with substrate was proportional to the amount of live cells following incubation with the sample. Potency results for mAb A, mAb B, and Pro X were expressed as percent relative potency to a nonlyophilized reference standard. Potency for mAb C was tested by measuring binding to its target receptor by the use of a molecule-specific surface plasmon resonance binding assay (Biacore; GE Healthcare Biosciences, Pittsburgh, Pennsylvania). Biological activity for mAb C was expressed as percent relative potency of the sample to a liquid, nonlyophilized, control.
Peptide Mapping Samples for peptide mapping analysis were denatured with guanidine hydrochloride, reduced with dithiothreitol, alkylated with iodoacetamide, and digested with trypsin. The digestion was quenched with trifluoroacetic acid, and the peptide fragments were analyzed using a Waters UPLCTM system coupled to an Orbitrap (ThermoElectron, Waltham, Massachusetts) mass spectrometer. The peptides were separated on a C18 column and monitored using a UV detector and a mass spectrometer in a positive ion mode.
Capillary Isoelectric Focusing Monoclonal antibody (mAb) B and Pro X were analyzed by capillary isoelectric focusing (cIEF) to determine any changes in charge variants. mAb B and Pro X samples were diluted to 2 and 8 mg/mL, respectively, with ultrapure water. mAb B was then digested with carboxypeptidase B for 10 min at 37 C and diluted with ultrapure water, 1% methyl cellulose, pharmalyte, and pI markers. Eight microliter of Pro X samples was mixed with 200 mL of a buffer master mix containing urea, ultrapure water, 1% methyl cellulose, pharmalyte, and pI markers. All samples were focused on an iCE280 or iCE3 (Protein Simple) for 1 and 10 min at 1500 and 3000 V, respectively.
Scanning Electron Microscopy Storage Stability Studies Scanning electron microscopy (SEM) was used to qualitatively assess cake morphology. Lyophilized cake samples were cut into large pieces and removed from the vial using a sharp blade. The pieces of cake were then spread onto white paper. The sharp blade was then used to slice the cake along the cylindrical axis; only the cake surface untouched by the blade was used for SEM examination by a bench top SEM instrument (Model TM-1000; Hitachi, Clarksburg, Maryland). All samples were sputter coated using gold with 10 nm thickness using a sputter coater from Electron Microscopy Sciences (Model EMS 150R ES, Hatfield, Pennsylvania). Potency In vitro potency of mAb A, mAb B, and Pro X was measured by an in-house cell-based assay specific for each protein. For mAb A and
Lyophilized samples were stored at 5 C, 25 C, 40 C, and 60 C to evaluate storage stability. At each time point, the product was visually inspected for cake appearance. Reconstitution was performed using water for injection (WFI), and reconstitution time was recorded as the duration between addition of WFI and observation of all solids dissolved in solution. Reconstituted samples were analyzed for purity by HPSEC (all proteins) and for SbVP by MFI (mAb C) to assess physical stability. Selected samples were also analyzed for potency, and chemical stability by cIEF and peptide mapping. Testing on the reconstituted samples occurred as soon as possible after reconstitution. To assess for any differences in physical stability of the reconstituted solution after the different drying conditions, reconstituted samples were also held at 40 C after the final 60 C time point of the respective lyophilized protein.
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
43
increasing the primary drying Ts led to significant decreases (>50%) in primary drying time (Table 1). 100 mg/mL mAb C Formulations Product temperatures during primary drying of mAb C at 100 mg/mL for the three different cycles are shown in Figure 2a. When Ts was 25 C (cycle 100C-1), the vials remained below both Tc and Tg0 . For cycles 100C-2 and 100C-3 (Ts of 0 C and þ30 C, respectively), vials were dried above Tg0 but remained below Tc. As shown in Figure 2a and Table 1, primary drying time decreased significantly as Ts increased. Approximately 80% reduction in primary drying time for mAb C was achieved when Ts increased from 25 C (41.4 h) to þ30 C (8.6 h). A similar result was found for mAb A; approximately 70% reduction in primary drying time was achieved when Ts increased from 25 C (43.5 h) to þ30 C (12.5 h) (Table 1). For both mAb A and mAb C, each of the three executed cycles yielded cakes that were structurally elegant with no evidence of collapse; no differences in cake appearances were evident among the three different cycles (not shown). Also, moisture contents were similar among all samples and were all less than 0.5%.
Figure 1. Glass transition temperature of the maximally freeze-concentrated solution and Tc as a function of protein concentration. Values for mAb A are denoted by triangles, mAb B by circles, mAb C by squares, and Pro X by diamonds. Closed symbols: Tg0 ; open symbols: Tc.
Results 25 mg/mL mAb C Formulations Thermal Analysis Product temperatures during primary drying of mAb C at 25 mg/ mL for the three different cycles are shown in Figure 2b. As shown in Figure 2b and Table 1, primary drying time decreased significantly as Ts increased. Approximately 87% reduction in primary drying time was achieved when Ts increased from 30 C (66.2 h) to þ15 C (8.8 h). When Ts was 30 C (cycle 25C-1), the vials appeared to remain below both Tc and Tg0 . For cycles 25C-2 and 25C3 (Ts of 0 C and þ15 C, respectively), vials were dried above Tc. Figure 3a shows representative pictures of the vials from each of the three cycles. The cakes from cycle 25C-1, which were below both Tc and Tg0 during primary drying, show no evidence of collapse. On the contrary, cakes from cycles 25C-2 and 25C-3 show areas of minor collapse throughout the batch, particularly near the bottom of the vial. Moisture contents were similar among all samples and were all less than 0.5%.
Differential scanning calorimetry measurements were performed to determine the glass transition temperature of the maximally Tg0 , and freeze-drying microscopy was used to determine the Tc. As shown in Figure 1, both Tg0 and Tc increased as protein concentration increased for both mAb A and mAb C. Results were similar between both molecules, which were formulated under identical buffer conditions. Additionally, consistent with results reported by Colandene et al.,11 Tc was generally higher than Tg0 for all proteins evaluated in this study, and the difference between Tc and Tg0 increased with concentration.
Lyophilization Cycle Data Monoclonal antibody (mAb) A was prepared at 100 mg/mL, and mAb C was prepared at three different concentrations (5, 25, and 100 mg/mL) for lyophilization. mAb B and Pro X were each prepared at 50 mg/mL for lyophilization. For each protein/concentration, two to three separate lyophilization cycles were executed that varied in Ts during primary drying as shown in Table 1. As expected,
5 mg/mL mAb C Formulations Product temperatures during primary drying of mAb C at 5 mg/mL for the three different cycles are shown in Figure 2c. As shown in Figure 2c and Table 1, primary drying time decreased
Table 1 Lyophilization of mAb A, mAb B, mAb C, and Pro X Protein Concentration (mg/mL)
Tg0 ( C)
Tc ( C)
DT ( C)
Cycle ID
1 Drying Ts ( C)
1 Drying Time (h)
100 mg/mL mAb A
21.9
9.0
12.9
50 mg/mL mAb B
27.0
16.0
11.0
100 mg/mL mAb C
22.0
8.0
14.0
25 mg/mL mAb C
27.5
24.5
3.0
5 mg/mL mAb C
32.4
30.0
2.4
50 mg/mL Pro X
25.0
20.0
5.0
100A-1 100A-2 100A-3 50B-1 50B-2 100C-1 100C-2 100C-3 25C-1 25C-2 25C-3 5C-1 5C-2 5C-3 50X-1 50X-2
25 0 þ30 25 5 25 0 þ30 30 0 þ15 36 30 15 20 0
43.5 16.6 12.5 55.7 26.8 41.4 14.0 8.6 66.2 14.8 8.8 106 44.4 17.4 57.6 19.6
DT refers to the difference in Tc and Tg0 (TceTg0 ) of the maximally freeze concentrated formulation. Cycle IDs were assigned to differentiate the primary drying Tss. Primary drying time was determined from comparative pressure measurements.
44
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
Figure 2. Product temperature data from the primary drying step of the lyophilization cycles executed for mAb C. Traces are from thermocouples inserted into different locations of the vial array: center (solid line), rear edge (dotted line), and front edge (dashed line), where front refers to the side closest to the door. Panels represent the three different antibody concentrations: (a) 100 mg/mL, (b) 25 mg/mL, and (c) 5 mg/mL. Each panel indicates each formulation's Tg0 and Tc as a horizontal line, and each set of three traces from a single cycle are identified by the cycle IDs referred to in Table 1.
significantly as Ts increased. Primary drying duration was almost 4.5 days when Ts was 36 C (cycle 5C-1). At this condition, primary drying occurred below both Tc and Tg0 . When Ts increased to 30 C (cycle 5C-2), primary drying time decreased significantly by approximately 58%. Primary drying occurred at or slightly below Tg0 for this cycle (Fig. 2c), and some areas of minor collapse were observed in most of the vials (Fig. 3b). An even further reduction in primary drying time was observed at a Ts of 15 C (cycle 5C-3, 17.4 h primary drying time), but this reduction was concomitant with a higher extent of cake collapse throughout the batch (Fig. 3b). Moisture contents were similar among all samples/cycles and were all 0.5%.
50 mg/mL mAb B and Pro X Formulations For both mAb B and Pro X, a cycle was executed wherein the product temperature remained below each formulation's Tg0 (50B-1
and 50X-1, respectively). In each case, primary drying was approximately 55 h. Each formulation was then freeze dried at a Ts in which the product temperature exceeded Tg0 but remained below the respective Tc (50B-2 and 50X-2). In each, the result was greater than 50% reduction in primary drying time (Table 1). Further, we observed no change to cake appearance and moisture content between the different drying conditions.
Cake Appearances by SEM Representative SEM images of mAb C cakes prepared from the 100-mg/mL formulations are shown in Figure 4a. No obvious visual differences in cake morphology were observed among the different primary drying conditions. A similar result was found for the 100mg/mL mAb A formulations and the 25-mg/mL mAb C formulations (not shown). For cakes prepared from the 5-mg/mL mAb C
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
45
Figure 3. Representative mAb C cakes after lyophilization of 25 mg/mL (a) and 5 mg/mL (b) formulations at different primary drying Tss. Areas of collapse are denoted by the arrows.
formulations, samples from cycle 5C-3 (Ts ¼ 15 C, dried above Tg0 per Fig. 2c) appear qualitatively different in morphology and pore size from samples from cycles 5C-1 and 5C-2 (Fig. 4b), which may be likely related to the higher extent of cake collapse noted in the previous section. Product temperature during primary drying was apparently below Tc in samples from cycle 5C-3 (Fig. 2c), which underscores the fact that lower concentration protein formulations may require a crystalline bulking agent in order to enable freezedrying near or above Tg0 to achieve acceptable cake structures. Stability of the Lyophilized Samples Lyophilized samples were placed on stability at multiple temperatures (5 C, 25 C, 40 C, and 60 C), and the proteins were evaluated for physical and chemical stability at various time points. Prior DSC measurements on the lyophilized solids showed that all Tg values exceeded 60 C, confirming the suitability of using 60 C as a storage condition while maintaining the cakes in the amorphous glassy state. Throughout the stability studies, no physical changes were observed in cake appearances at any of the storage temperatures from what was observed immediately upon lyophilization. Further, no changes were observed in reconstitution time during storage from what was measured at the initial time point. For all mAb A and mAb C cakes lyophilized from formulations at 100 mg/mL, reconstitution time was generally 30e60 min at all storage time and temperature conditions. General reconstitution times for the mAb C formulations at 25 and 5 mg/mL were less than 5 min and less than 1 min, respectively. The increase in reconstitution time with increasing protein concentration was expected based on other reported results, particularly the relatively long times noted at 100 mg/mL.23-25 Moreover, it is worth noting that for each particular
protein/concentration in this study, no differences in reconstitution time were evident among the different drying conditions. Physical Stability by HPSEC The primary degradation pathway for each protein in the lyophilized solid was aggregation as determined by HPSEC. Representative stability data are presented in Figure 5, which show HPSEC monomer purity profiles during storage at 40 C for mAb A lyophilized at 100 mg/mL and mAb C lyophilized at 5, 25, and 100 mg/mL. Table 2 shows aggregation rates at all temperatures for the different formulations prepared in this study. The data show that for each protein/concentration condition, the stability of the lyophilized protein was not affected by the primary drying condition. To investigate whether any stability differences by HPSEC could be observed during longer holds after reconstitution, reconstituted samples were stored at 40 C for either 1 month (mAb A) or 3 months (mAb C) at the end of each respective formulation's hold at 60 C in the lyophilized solid. The data in Figure 6 show that the physical stability of the formulations was not impacted by the drying condition upon reconstitution and holding at 40 C as determined by HPSEC. Physical Stability by SbVP Analysis Subvisible particle analysis was performed to measure any aggregates that may not be detected by HPSEC.26 For mAb C, SbVP data as determined by MFI for particle sizes 2 mm are shown in Figure 7 for the formulations prepared at 5, 25, and 100 mg/mL. The data show no obvious effect of drying condition on SbVP levels throughout storage at any of the studied storage temperatures.
46
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
Figure 4. Scanning electron microscopy images of representative cakes after lyophilization of 100 mg/mL (a) and 5 mg/mL (b) at different primary drying Tss.
Similar findings were observed at the size ranges 10 and 25 mm (data not shown). Also, using light obscuration analysis, no impact of drying condition was observed on SbVP levels during storage of mAb A, mAb B, and Pro X samples (data not shown).
mAb B (50 mg/mL), mAb C (5 and 100 mg/mL), and Pro X (50 mg/ mL) show no significant difference in relative potency after longterm storage between drying cycles executed at the respective lowest and highest primary drying Tss.
Potency
Chemical Stability by cIEF
In vitro potency measurements show no impact of drying conditions (Table 3). Bioactivity measurements for mAb A (100 mg/mL),
Chemical stability of mAb B and Pro X was assessed by determining charge variants by cIEF. Each of these proteins was freeze
▪
Figure 5. Monomer content by HPSEC for lyophilized mAbs A and C during storage at 40 C comparing three different primary drying Tss. (,) Lowest Tss, ( ) intermediate Ts, and (:) highest Ts for each mAb concentration as listed in Table 1.
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
47
Table 2 Aggregation Rates in the Dried Solid for mAb A, mAb B, mAb C, and Pro X During Storage at Different Temperatures After Drying at the Different Conditions Referred to in Table 1 mAb Concentration (mg/mL)
100 mg/mL mAb A
50 mg/mL mAb B 100 mg/mL mAb C
25 mg/mL mAb C
5 mg/mL mAb C
50 mg/mL Pro X
Cycle ID
Storage Temperature
100A-1 100A-2 100A-3 50B-1 50B-2 100C-1 100C-2 100C-3 25C-1 25C-2 25C-3 5C-1 5C-2 5C-3 50X-1 50X-2
5 C (24 M)
25 C (6 M)
40 C (3 M)
60 C (7d)
0.14 0.11 0.19 0.48 (6M) 0.40 (6M) 0.17 0.17 0.14 0.010 0.011 0.019 0.036 0.046 0.033 0.32 (9M)a 0.0010 (9M)
0.19 0.18 0.21 0.17 0.21 0.26 0.25 0.25 0.010 0.010 0.012 0.041 0.043 0.041 0.026 0.0071
1.2 1.2 1.2 0.77 0.98 1.3 1.3 1.3 0.048 0.041 0.043 0.038 0.040 0.014 0.13 0.043
0.47 0.37 0.46 ND ND 0.42 0.45 0.46 0.030 0.037 0.036 0.044 0.050 0.032 ND ND
Aggregation rates for storage at 5 C, 25 C, 40 C, and 60 C are shown as percentage aggregates per year, month, and day, respectively. The number in parentheses next to the temperature in the heading is the duration of storage in the dried solid unless otherwise noted in the table (M, months; d, days). ND, not determined. a HPSEC variability and relative stability of Pro X at 5 C leads to apparent negative rate for aggregation.
dried at a Ts in which the product temperature exceeded Tg0 but remained below the respective Tc (50B-2 and 50X-2; Table 1). cIEF measurements obtained upon lyophilization and after 6 months at 40 C indicated no change in charge variants during storage in these samples that were dried above Tg0 (Table 4). Chemical Stability by Peptide Mapping Peptide mapping analysis was conducted to measure any chemical degradation products during long-term storage of mAb A and mAb C. For mAb C lyophilized at 100 mg/mL at the lowest and highest primary drying Tss (100C-1 and 100C-3 in Table 1), peptide mapping analysis conducted after 24 months storage at 5 C
revealed low levels of chemical variants in both samples with no significant differences between them. Deamidation and methionine oxidation levels for susceptible residues were less than 5%. Similarly, for mAb C lyophilized at 5 mg/mL at the lowest and highest primary drying Tss (5C-1 and 5C-3 in Table 1), deamidation and methionine oxidation levels for individual susceptible residues were less than 12% with no significant differences between the two samples. For mAb A, the only observed chemical modification was deamidation, and deamidation levels in samples dried below and above Tg0 (100A-1 and 100A-3; Table 1) remained consistent (
Contents lists available at ScienceDirect
Journal of Pharmaceutical Sciences journal homepage: www.jpharmsci.org
Pharmaceutical Biotechnology
Freeze-Drying Above the Glass Transition Temperature in Amorphous Protein Formulations While Maintaining Product Quality and Improving Process Efficiency Roberto A. Depaz, Swapnil Pansare, Sajal Manubhai Patel* Department of Formulation Sciences, Biopharmaceutical Development, MedImmune, 1 MedImmune Way, 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 3 June 2015 Revised 31 August 2015 Accepted 5 October 2015 Available online 18 November 2015
This study explored the ability to conduct primary drying during lyophilization at product temperatures above the glass transition temperature of the maximally freeze-concentrated solution (Tg0 ) in amorphous formulations for four proteins from three different classes. Drying above Tg0 resulted in significant reductions in lyophilization cycle time. At higher protein concentrations, formulations freeze dried above Tg0 but below the collapse temperature yielded pharmaceutically acceptable cakes. However, using an immunoglobulin G type 4 monoclonal antibody as an example, we found that as protein concentration decreased, minor extents of collapse were observed in formulations dried at higher temperatures. No other impacts to product quality, physical stability, or chemical stability were observed in this study among the different drying conditions for the different proteins. Drying amorphous formulations above Tg0 , particularly high protein concentration formulations, is a viable means to achieve significant time and cost savings in freeze-drying processes. © 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
Keywords: freeze-drying lyophilization protein formulation thermal analysis glass transition stability processing drying
Introduction Freeze-drying, or lyophilization, is a unit operation commonly used to stabilize protein formulations to achieve longer shelf life, compared with solution formulations, during product development and commercial distribution. In the last decade, more than 40% of marketed biotherapeutics were freeze dried,1 and with new complex molecular formats being developed (e.g., bispecific molecules, fusion proteins, antibody drug conjugates), this percentage is expected to further increase. However, the processes of freezing and drying could each be detrimental to protein stability. Thus, optimization of the formulation composition so that the protein is stable against the stresses imposed by freeze-drying is a key element of drug product development.2-4 Additionally, optimization of the freeze-drying process is equally critical to minimize
Abbreviations used: HPSEC, high-performance size-exclusion chromatography; IgG1, immunoglobulin G type 1; IgG4, immunoglobulin G type 4; mAb, monoclonal antibody; MFI, microflow imaging; SbVP, subvisible particle; Tc, collapse temperature determined by freeze-dry microscopy; Tg0 , glass transition temperature of the maximally freeze-concentrated solution; Ts, shelf temperature. * Correspondence to: Sajal M. Patel (Telephone: 301-398-5247; Fax: 301-3987782). E-mail address: [email protected] (S.M. Patel).
processing time for increased throughput and also to reduce production costs.5 A typical freeze-drying process consists of three steps: (1) freezing wherein water is converted into ice; (2) primary drying wherein ice is removed by sublimation; and (3) secondary drying wherein unfrozen water is removed by desorption. Typically, at the end of secondary drying the residual water content is less than 1%. Of these three steps, primary drying is often the longest step and hence optimization of this step is usually the main focus in industry. During primary drying, the product temperature needs to be maintained 2 Ce3 C below the maximum allowable temperature, which for an amorphous matrix is the glass transition temperature of maximally freeze-concentrated solution (Tg0 ) or the collapse temperature (Tc).6 Drying above Tg0 or Tc may result in cake collapse that could further impact product quality attributes such as residual water content, reconstitution time, and protein stability.7-9 Usually, Tc is within 1 Ce2 C of Tg0 , and Tg0 and Tc may be used interchangeably.10 However, for higher concentration protein formulations, visible collapse may not always be observed when drying above Tg0 as the difference between Tg0 and the Tc may increase as protein concentration increases.11 Recent publications have shown that even with total collapse, protein stability either improves or is not significantly different compared with that in the uncollapsed cake.12-14 However, cake
http://dx.doi.org/10.1002/jps.24705 0022-3549/© 2016 American Pharmacists Association®. Published by Elsevier Inc. All rights reserved.
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
41
and stoppers were obtained from West Pharmaceutical Services, Inc. (Exton, Pennsylvania).
appearance is a potential critical quality attribute for a lyophilized product and hence included in product specifications for lot release and stability. In general, improvement in process efficiency at the expense of product appearance may not be desired. Thus, although drying above the Tc may not be practical from an appearance perspective, drying above Tg0 is a very practical approach to reduce freeze-drying cycle time without impacting product quality,11 especially for high concentration protein formulations. Furthermore, for existing processes where primary drying occurs at product temperatures below Tg0 , data supporting maintained product quality if product temperature exceed Tg0 during primary drying is invaluable in the event of any temperature and pressure excursions that may occur during production. On the basis of the Fox equation and the relatively high Tg0 values of proteins, the Tg0 of a protein-containing amorphous formulation is expected to increase with protein concentration assuming an otherwise constant formulation composition.15,16 However, understanding the effect that increasing protein concentration may have on Tc, and the extent to which Tc may differ from Tg0 , is critical in defining the maximum allowable product temperature during primary drying. Because every 1 C increase in product temperature during primary drying may result in approximately 13% reduction in primary drying time,6 operating at the highest allowable product temperature is highly desired. Thus, thermal characterization of the formulation by differential scanning calorimetry (DSC) for Tg0 and freeze-dry microscopy (FDM) for Tc is critical for both product quality and process efficiency. In this work, both product quality and process performance were evaluated when conducting primary drying at product temperatures above and below Tg0 for four different proteins [two immunoglobulin G type 1 (IgG1) monoclonal antibodies, an immunoglobulin G type 4 (IgG4) monoclonal antibody (mAb), and a fusion protein] in an amorphous formulation matrix. Different primary drying conditions were systematically evaluated, and stability studies were performed to show there was no impact to product quality both upon lyophilization and also during long-term storage. The relationship between Tg0 and Tc as a function of protein concentration is presented for the various molecules, indicating the potential to exceed product temperature above Tg0 during primary drying, particularly at higher protein concentrations, to achieve significant reduction in lyophilization cycle time.
Vials were partially stoppered with 13 mm (for 3-mL vials) or 20 mm (for 5-mL and 20-mL vials) lyophilization stoppers. Freezedrying was performed using a Virtis Genesis 35 EL freeze dryer (SP Scientific, Stone Ridge, New York) or Millrock PDQ24XS-S freeze dryer (Millrock Technology, Kingston, New York). Vials were loaded onto the freeze dryer at a shelf temperature (Ts) of 20 C. After loading the shelves, they were cooled to 5 C and held for 30 min. The shelves were then cooled to 5 C and held for 15 min before cooling to 40 C. All molecules except mAb C underwent an annealing step at 16 C for 120 min before bringing back to 40 C. The annealing time and temperature was selected based on process optimization activities leading to batch homogeneity in terms of drying and product quality. After holding at 40 for 2 h, the Ts was raised to the desired primary drying Ts (see Results section) where primary drying occurred at a chamber pressure of 100 mTorr. Chamber pressure was controlled using the Capacitance Manometer gauge. End of primary drying was determined by comparative pressure measurement wherein the Pirani gauge pressure measurement converges with that of the Capacitance Manometer chamber pressure.17 Secondary drying was performed at 40 C for 6 h while maintaining the pressure at 100 mTorr. Product temperature was monitored during the freeze-drying process using thermocouples placed at the bottom center of selected vials. The variability in product temperature as measured by the thermocouples was ±0.5 C.18 With the following two exceptions, all Ts ramps throughout the cycle were conducted at 0.5 C/min: (1) for mAb A, the ramp rate from 40 C to the primary drying Ts was 0.1 C/min, and (2) all ramp rates during secondary drying to 40 C were 0.1 C/min.
Materials and Methods
Differential Scanning Calorimetry
Materials
A Q2000 series differential scanning calorimeter from TA Instruments (New Castle, Delaware) was used for glass transition temperature measurements. For Tg0 measurements, 20 mL of liquid sample was added into an aluminum pan and sealed hermetically. An empty pan and lid was used as the reference. Liquid samples were frozen to 60 C at a rate of 5 C/min and then heated to 25 C at a rate of 5 C/min. The Tg0 values were determined using Universal Analysis software and reported as the midpoint of the glass transition.19-21 For Tg determination of the lyophilized cakes, approximately 5 mg of sample were sealed into an aluminum pan and analyzed in modulated DSC mode. Pans were heated at 5 C/min through the glass transition with a modulation period of 80 s and amplitude ±0.5 C.
The proteins evaluated in this work were produced and purified at MedImmune (Gaithersburg, Maryland) using proprietary methods. mAb A and mAb B are IgG1 monoclonal antibodies, and mAb C is an IgG4 mAb. Protein X (or “Pro X”) is a fusion protein with a molecular weight of approximately 90 kDa. Each of these molecules was formulated in a buffered solution with a disaccharide and polysorbate 80 as excipients. mAb B contained an amino acid as an additional excipient. Formulations contained excipients expected to remain amorphous upon lyophilization, and no components expected to crystallize were included. All excipients used were multicompendial grade. For thermal analysis, mAb A and mAb C samples were prepared in their formulation buffer at concentrations ranging between 1 and 100 mg/mL. For lyophilization, mAb A was prepared at 100 mg/mL, and mAb C was prepared at three different concentrations: 5, 25, and 100 mg/mL with the levels of all other formulation components remaining the same. mAb B and Pro X were prepared at 50 mg/mL for thermal analysis and lyophilization. Type 1 glass tubing vials (3-, 5-, and 20-mL), and chlorobutyl, single vent lyophilization stoppers were used. Both vials
Lyophilization For their individual lyophilization cycles, the proteins were filled into vials at the following conditions: mAb A and mAb C: 3-mL vial, 1.1 mL fill mAb B: 5-mL vial, 2.7-mL fill Pro X: 20-mL vial, 5.5-mL fill
Freeze-Dry Microscopy All liquid samples were tested using an Olympus BX50 microscope with a Linkam FDCS 196 stage. Images were recorded using a QImaging camera attachment. A sample volume of 5-10 mL was placed between two cover slips on the FDM stage. The samples were cooled and frozen to 40 C. All samples, with the exception of
42
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
mAb C, then underwent an annealing step at 16 C followed by refreeze to 40 C. A rate of 10 C/min was used throughout the freezing steps. After freezing, a vacuum at 100 mTorr was initiated using a vacuum pump, and heating was initiated at a rate of 1 C/min. The ramp rate was decreased to 0.5 C/min when the temperature exceeded Tg0 and was near the Tc. Onset of collapse was recorded as the Tc, when a change in the structure of the freezedried solid was observed as described elsewhere.22 High-Performance Size-Exclusion Chromatography High-performance size-exclusion chromatography (HPSEC) was performed using an Agilent HPLC system with UV absorbance at 280 nm. A 7.8 300 mm2 TSKgel G3000SWXL column with 6.0 40 mm2 TSKgel SWXL guard column (Tosoh Bio-science, King of Prussia, Pennsylvania) was used with a mobile phase comprising 0.1 M sodium phosphate, 0.1 M sodium sulfate, 0.05% (w/v) sodium azide, pH 6.8 (the mobile phase for Pro X was 90% of this solution along with 10% isopropanol). Formulations at 25, 50, and 100 mg/mL were diluted to 10 mg/mL with phosphate-buffered saline (Gibco Life Technologies Corporation, Grand Island, New York) prior to analysis, and 25 mL were injected. Formulations at 5 mg/mL were injected neat with an injection volume of 50 mL. The flow rate was 1 mL/min, and samples were held at 5 C in the autosampler tray prior to injection. The relative percentage of each species in each sample (aggregates, monomer, fragments) was calculated relative to the total area of all peaks. Moisture Determination Residual moisture content was determined using a Mettler Toledo DL39 Karl Fischer (KF) coulometer. Lyophilized product was reconstituted with anhydrous methanol and then injected into the KF coulometer for analysis. The percentage residual moisture content was determined based upon the amount of water titrated relative to the mass of the lyophilized sample. Microflow Imaging Subvisible particle (SbVP) analysis by microflow imaging (MFI) was performed using an MFI 5200 from Protein Simple (Santa Clara, California). Samples were analyzed neat. Prior to each sample measurement, the background illumination was optimized using formulation buffer. Approximately 0.25 mL sample was allowed to flow through prior to counting to generate a background free of Schlieren effects. Counts for particle sizes 2 mm, 10 mm, and 25 mm were recorded for particles with an aspect ratio less than 0.85.
Pro X, the bioassay measured the ability of the protein to inhibit binding of the respective molecule target to cells expressing a luciferase reporter gene. Luciferase expression was quantified via the use of a chemiluminescent substrate and an Envision plate reader (Perkin Elmer, Waltham, MA). For mAb B, the bioassay measured the ability of the protein to protect cells from cell death. The amount of luminescence measured after reaction with substrate was proportional to the amount of live cells following incubation with the sample. Potency results for mAb A, mAb B, and Pro X were expressed as percent relative potency to a nonlyophilized reference standard. Potency for mAb C was tested by measuring binding to its target receptor by the use of a molecule-specific surface plasmon resonance binding assay (Biacore; GE Healthcare Biosciences, Pittsburgh, Pennsylvania). Biological activity for mAb C was expressed as percent relative potency of the sample to a liquid, nonlyophilized, control.
Peptide Mapping Samples for peptide mapping analysis were denatured with guanidine hydrochloride, reduced with dithiothreitol, alkylated with iodoacetamide, and digested with trypsin. The digestion was quenched with trifluoroacetic acid, and the peptide fragments were analyzed using a Waters UPLCTM system coupled to an Orbitrap (ThermoElectron, Waltham, Massachusetts) mass spectrometer. The peptides were separated on a C18 column and monitored using a UV detector and a mass spectrometer in a positive ion mode.
Capillary Isoelectric Focusing Monoclonal antibody (mAb) B and Pro X were analyzed by capillary isoelectric focusing (cIEF) to determine any changes in charge variants. mAb B and Pro X samples were diluted to 2 and 8 mg/mL, respectively, with ultrapure water. mAb B was then digested with carboxypeptidase B for 10 min at 37 C and diluted with ultrapure water, 1% methyl cellulose, pharmalyte, and pI markers. Eight microliter of Pro X samples was mixed with 200 mL of a buffer master mix containing urea, ultrapure water, 1% methyl cellulose, pharmalyte, and pI markers. All samples were focused on an iCE280 or iCE3 (Protein Simple) for 1 and 10 min at 1500 and 3000 V, respectively.
Scanning Electron Microscopy Storage Stability Studies Scanning electron microscopy (SEM) was used to qualitatively assess cake morphology. Lyophilized cake samples were cut into large pieces and removed from the vial using a sharp blade. The pieces of cake were then spread onto white paper. The sharp blade was then used to slice the cake along the cylindrical axis; only the cake surface untouched by the blade was used for SEM examination by a bench top SEM instrument (Model TM-1000; Hitachi, Clarksburg, Maryland). All samples were sputter coated using gold with 10 nm thickness using a sputter coater from Electron Microscopy Sciences (Model EMS 150R ES, Hatfield, Pennsylvania). Potency In vitro potency of mAb A, mAb B, and Pro X was measured by an in-house cell-based assay specific for each protein. For mAb A and
Lyophilized samples were stored at 5 C, 25 C, 40 C, and 60 C to evaluate storage stability. At each time point, the product was visually inspected for cake appearance. Reconstitution was performed using water for injection (WFI), and reconstitution time was recorded as the duration between addition of WFI and observation of all solids dissolved in solution. Reconstituted samples were analyzed for purity by HPSEC (all proteins) and for SbVP by MFI (mAb C) to assess physical stability. Selected samples were also analyzed for potency, and chemical stability by cIEF and peptide mapping. Testing on the reconstituted samples occurred as soon as possible after reconstitution. To assess for any differences in physical stability of the reconstituted solution after the different drying conditions, reconstituted samples were also held at 40 C after the final 60 C time point of the respective lyophilized protein.
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
43
increasing the primary drying Ts led to significant decreases (>50%) in primary drying time (Table 1). 100 mg/mL mAb C Formulations Product temperatures during primary drying of mAb C at 100 mg/mL for the three different cycles are shown in Figure 2a. When Ts was 25 C (cycle 100C-1), the vials remained below both Tc and Tg0 . For cycles 100C-2 and 100C-3 (Ts of 0 C and þ30 C, respectively), vials were dried above Tg0 but remained below Tc. As shown in Figure 2a and Table 1, primary drying time decreased significantly as Ts increased. Approximately 80% reduction in primary drying time for mAb C was achieved when Ts increased from 25 C (41.4 h) to þ30 C (8.6 h). A similar result was found for mAb A; approximately 70% reduction in primary drying time was achieved when Ts increased from 25 C (43.5 h) to þ30 C (12.5 h) (Table 1). For both mAb A and mAb C, each of the three executed cycles yielded cakes that were structurally elegant with no evidence of collapse; no differences in cake appearances were evident among the three different cycles (not shown). Also, moisture contents were similar among all samples and were all less than 0.5%.
Figure 1. Glass transition temperature of the maximally freeze-concentrated solution and Tc as a function of protein concentration. Values for mAb A are denoted by triangles, mAb B by circles, mAb C by squares, and Pro X by diamonds. Closed symbols: Tg0 ; open symbols: Tc.
Results 25 mg/mL mAb C Formulations Thermal Analysis Product temperatures during primary drying of mAb C at 25 mg/ mL for the three different cycles are shown in Figure 2b. As shown in Figure 2b and Table 1, primary drying time decreased significantly as Ts increased. Approximately 87% reduction in primary drying time was achieved when Ts increased from 30 C (66.2 h) to þ15 C (8.8 h). When Ts was 30 C (cycle 25C-1), the vials appeared to remain below both Tc and Tg0 . For cycles 25C-2 and 25C3 (Ts of 0 C and þ15 C, respectively), vials were dried above Tc. Figure 3a shows representative pictures of the vials from each of the three cycles. The cakes from cycle 25C-1, which were below both Tc and Tg0 during primary drying, show no evidence of collapse. On the contrary, cakes from cycles 25C-2 and 25C-3 show areas of minor collapse throughout the batch, particularly near the bottom of the vial. Moisture contents were similar among all samples and were all less than 0.5%.
Differential scanning calorimetry measurements were performed to determine the glass transition temperature of the maximally Tg0 , and freeze-drying microscopy was used to determine the Tc. As shown in Figure 1, both Tg0 and Tc increased as protein concentration increased for both mAb A and mAb C. Results were similar between both molecules, which were formulated under identical buffer conditions. Additionally, consistent with results reported by Colandene et al.,11 Tc was generally higher than Tg0 for all proteins evaluated in this study, and the difference between Tc and Tg0 increased with concentration.
Lyophilization Cycle Data Monoclonal antibody (mAb) A was prepared at 100 mg/mL, and mAb C was prepared at three different concentrations (5, 25, and 100 mg/mL) for lyophilization. mAb B and Pro X were each prepared at 50 mg/mL for lyophilization. For each protein/concentration, two to three separate lyophilization cycles were executed that varied in Ts during primary drying as shown in Table 1. As expected,
5 mg/mL mAb C Formulations Product temperatures during primary drying of mAb C at 5 mg/mL for the three different cycles are shown in Figure 2c. As shown in Figure 2c and Table 1, primary drying time decreased
Table 1 Lyophilization of mAb A, mAb B, mAb C, and Pro X Protein Concentration (mg/mL)
Tg0 ( C)
Tc ( C)
DT ( C)
Cycle ID
1 Drying Ts ( C)
1 Drying Time (h)
100 mg/mL mAb A
21.9
9.0
12.9
50 mg/mL mAb B
27.0
16.0
11.0
100 mg/mL mAb C
22.0
8.0
14.0
25 mg/mL mAb C
27.5
24.5
3.0
5 mg/mL mAb C
32.4
30.0
2.4
50 mg/mL Pro X
25.0
20.0
5.0
100A-1 100A-2 100A-3 50B-1 50B-2 100C-1 100C-2 100C-3 25C-1 25C-2 25C-3 5C-1 5C-2 5C-3 50X-1 50X-2
25 0 þ30 25 5 25 0 þ30 30 0 þ15 36 30 15 20 0
43.5 16.6 12.5 55.7 26.8 41.4 14.0 8.6 66.2 14.8 8.8 106 44.4 17.4 57.6 19.6
DT refers to the difference in Tc and Tg0 (TceTg0 ) of the maximally freeze concentrated formulation. Cycle IDs were assigned to differentiate the primary drying Tss. Primary drying time was determined from comparative pressure measurements.
44
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
Figure 2. Product temperature data from the primary drying step of the lyophilization cycles executed for mAb C. Traces are from thermocouples inserted into different locations of the vial array: center (solid line), rear edge (dotted line), and front edge (dashed line), where front refers to the side closest to the door. Panels represent the three different antibody concentrations: (a) 100 mg/mL, (b) 25 mg/mL, and (c) 5 mg/mL. Each panel indicates each formulation's Tg0 and Tc as a horizontal line, and each set of three traces from a single cycle are identified by the cycle IDs referred to in Table 1.
significantly as Ts increased. Primary drying duration was almost 4.5 days when Ts was 36 C (cycle 5C-1). At this condition, primary drying occurred below both Tc and Tg0 . When Ts increased to 30 C (cycle 5C-2), primary drying time decreased significantly by approximately 58%. Primary drying occurred at or slightly below Tg0 for this cycle (Fig. 2c), and some areas of minor collapse were observed in most of the vials (Fig. 3b). An even further reduction in primary drying time was observed at a Ts of 15 C (cycle 5C-3, 17.4 h primary drying time), but this reduction was concomitant with a higher extent of cake collapse throughout the batch (Fig. 3b). Moisture contents were similar among all samples/cycles and were all 0.5%.
50 mg/mL mAb B and Pro X Formulations For both mAb B and Pro X, a cycle was executed wherein the product temperature remained below each formulation's Tg0 (50B-1
and 50X-1, respectively). In each case, primary drying was approximately 55 h. Each formulation was then freeze dried at a Ts in which the product temperature exceeded Tg0 but remained below the respective Tc (50B-2 and 50X-2). In each, the result was greater than 50% reduction in primary drying time (Table 1). Further, we observed no change to cake appearance and moisture content between the different drying conditions.
Cake Appearances by SEM Representative SEM images of mAb C cakes prepared from the 100-mg/mL formulations are shown in Figure 4a. No obvious visual differences in cake morphology were observed among the different primary drying conditions. A similar result was found for the 100mg/mL mAb A formulations and the 25-mg/mL mAb C formulations (not shown). For cakes prepared from the 5-mg/mL mAb C
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
45
Figure 3. Representative mAb C cakes after lyophilization of 25 mg/mL (a) and 5 mg/mL (b) formulations at different primary drying Tss. Areas of collapse are denoted by the arrows.
formulations, samples from cycle 5C-3 (Ts ¼ 15 C, dried above Tg0 per Fig. 2c) appear qualitatively different in morphology and pore size from samples from cycles 5C-1 and 5C-2 (Fig. 4b), which may be likely related to the higher extent of cake collapse noted in the previous section. Product temperature during primary drying was apparently below Tc in samples from cycle 5C-3 (Fig. 2c), which underscores the fact that lower concentration protein formulations may require a crystalline bulking agent in order to enable freezedrying near or above Tg0 to achieve acceptable cake structures. Stability of the Lyophilized Samples Lyophilized samples were placed on stability at multiple temperatures (5 C, 25 C, 40 C, and 60 C), and the proteins were evaluated for physical and chemical stability at various time points. Prior DSC measurements on the lyophilized solids showed that all Tg values exceeded 60 C, confirming the suitability of using 60 C as a storage condition while maintaining the cakes in the amorphous glassy state. Throughout the stability studies, no physical changes were observed in cake appearances at any of the storage temperatures from what was observed immediately upon lyophilization. Further, no changes were observed in reconstitution time during storage from what was measured at the initial time point. For all mAb A and mAb C cakes lyophilized from formulations at 100 mg/mL, reconstitution time was generally 30e60 min at all storage time and temperature conditions. General reconstitution times for the mAb C formulations at 25 and 5 mg/mL were less than 5 min and less than 1 min, respectively. The increase in reconstitution time with increasing protein concentration was expected based on other reported results, particularly the relatively long times noted at 100 mg/mL.23-25 Moreover, it is worth noting that for each particular
protein/concentration in this study, no differences in reconstitution time were evident among the different drying conditions. Physical Stability by HPSEC The primary degradation pathway for each protein in the lyophilized solid was aggregation as determined by HPSEC. Representative stability data are presented in Figure 5, which show HPSEC monomer purity profiles during storage at 40 C for mAb A lyophilized at 100 mg/mL and mAb C lyophilized at 5, 25, and 100 mg/mL. Table 2 shows aggregation rates at all temperatures for the different formulations prepared in this study. The data show that for each protein/concentration condition, the stability of the lyophilized protein was not affected by the primary drying condition. To investigate whether any stability differences by HPSEC could be observed during longer holds after reconstitution, reconstituted samples were stored at 40 C for either 1 month (mAb A) or 3 months (mAb C) at the end of each respective formulation's hold at 60 C in the lyophilized solid. The data in Figure 6 show that the physical stability of the formulations was not impacted by the drying condition upon reconstitution and holding at 40 C as determined by HPSEC. Physical Stability by SbVP Analysis Subvisible particle analysis was performed to measure any aggregates that may not be detected by HPSEC.26 For mAb C, SbVP data as determined by MFI for particle sizes 2 mm are shown in Figure 7 for the formulations prepared at 5, 25, and 100 mg/mL. The data show no obvious effect of drying condition on SbVP levels throughout storage at any of the studied storage temperatures.
46
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
Figure 4. Scanning electron microscopy images of representative cakes after lyophilization of 100 mg/mL (a) and 5 mg/mL (b) at different primary drying Tss.
Similar findings were observed at the size ranges 10 and 25 mm (data not shown). Also, using light obscuration analysis, no impact of drying condition was observed on SbVP levels during storage of mAb A, mAb B, and Pro X samples (data not shown).
mAb B (50 mg/mL), mAb C (5 and 100 mg/mL), and Pro X (50 mg/ mL) show no significant difference in relative potency after longterm storage between drying cycles executed at the respective lowest and highest primary drying Tss.
Potency
Chemical Stability by cIEF
In vitro potency measurements show no impact of drying conditions (Table 3). Bioactivity measurements for mAb A (100 mg/mL),
Chemical stability of mAb B and Pro X was assessed by determining charge variants by cIEF. Each of these proteins was freeze
▪
Figure 5. Monomer content by HPSEC for lyophilized mAbs A and C during storage at 40 C comparing three different primary drying Tss. (,) Lowest Tss, ( ) intermediate Ts, and (:) highest Ts for each mAb concentration as listed in Table 1.
R.A. Depaz et al. / Journal of Pharmaceutical Sciences 105 (2016) 40e49
47
Table 2 Aggregation Rates in the Dried Solid for mAb A, mAb B, mAb C, and Pro X During Storage at Different Temperatures After Drying at the Different Conditions Referred to in Table 1 mAb Concentration (mg/mL)
100 mg/mL mAb A
50 mg/mL mAb B 100 mg/mL mAb C
25 mg/mL mAb C
5 mg/mL mAb C
50 mg/mL Pro X
Cycle ID
Storage Temperature
100A-1 100A-2 100A-3 50B-1 50B-2 100C-1 100C-2 100C-3 25C-1 25C-2 25C-3 5C-1 5C-2 5C-3 50X-1 50X-2
5 C (24 M)
25 C (6 M)
40 C (3 M)
60 C (7d)
0.14 0.11 0.19 0.48 (6M) 0.40 (6M) 0.17 0.17 0.14 0.010 0.011 0.019 0.036 0.046 0.033 0.32 (9M)a 0.0010 (9M)
0.19 0.18 0.21 0.17 0.21 0.26 0.25 0.25 0.010 0.010 0.012 0.041 0.043 0.041 0.026 0.0071
1.2 1.2 1.2 0.77 0.98 1.3 1.3 1.3 0.048 0.041 0.043 0.038 0.040 0.014 0.13 0.043
0.47 0.37 0.46 ND ND 0.42 0.45 0.46 0.030 0.037 0.036 0.044 0.050 0.032 ND ND
Aggregation rates for storage at 5 C, 25 C, 40 C, and 60 C are shown as percentage aggregates per year, month, and day, respectively. The number in parentheses next to the temperature in the heading is the duration of storage in the dried solid unless otherwise noted in the table (M, months; d, days). ND, not determined. a HPSEC variability and relative stability of Pro X at 5 C leads to apparent negative rate for aggregation.
dried at a Ts in which the product temperature exceeded Tg0 but remained below the respective Tc (50B-2 and 50X-2; Table 1). cIEF measurements obtained upon lyophilization and after 6 months at 40 C indicated no change in charge variants during storage in these samples that were dried above Tg0 (Table 4). Chemical Stability by Peptide Mapping Peptide mapping analysis was conducted to measure any chemical degradation products during long-term storage of mAb A and mAb C. For mAb C lyophilized at 100 mg/mL at the lowest and highest primary drying Tss (100C-1 and 100C-3 in Table 1), peptide mapping analysis conducted after 24 months storage at 5 C
revealed low levels of chemical variants in both samples with no significant differences between them. Deamidation and methionine oxidation levels for susceptible residues were less than 5%. Similarly, for mAb C lyophilized at 5 mg/mL at the lowest and highest primary drying Tss (5C-1 and 5C-3 in Table 1), deamidation and methionine oxidation levels for individual susceptible residues were less than 12% with no significant differences between the two samples. For mAb A, the only observed chemical modification was deamidation, and deamidation levels in samples dried below and above Tg0 (100A-1 and 100A-3; Table 1) remained consistent (
