Peer-Reviewed Technical Note: Quality by Design in Freeze-Drying

The pharmaceutical industry and the US Food and Drug Administration have come to realize that testing quality into a final product can inhibit the rate of introducing new drugs to the market. Since 2003, FDA has been working to modernize pharmaceutical manufacturing and encouraging the use of process analytical technology (PAT) and quality by design (QbD) under its Pharmaceutical CGMPs for the 21st Century initiative (1). The goal of the initiative is to enable manufacturers to produce drug products more efficiently and with higher quality. To achieve this goal, manufacturers must specify critical process parameters (CPPs) and control them within predefined limits of the so-called "design space" (2). These specifications, however, must be developed with laboratory experiments using PAT tools that have the capability to measure CPPs with acceptable accuracy (3). With freeze-drying, in particular, the use of PAT in the laboratory is more focused on process understanding, whereas PAT in production-scale freeze-drying is necessary for process control and may expedite scale-up. The "SMART" freeze-dryer concept (FTS Systems, Stone Ridge, NY) which is based on manometric temperature measurement (MTM), is a new PAT tool that allows accurate product temperature determination at the ice sublimation interface (T_p) and the collection of product resistance data (R_p) during primary drying (4–6). In addition, SMART technology provides an optimized cycle recipe during the first laboratory experiment for a given formulation. This article describes an approach to evaluate the robustness of a product based on a few simple laboratory experiments for the purposes of scaling-up and to study the impact of a potential T_p excess during primary drying.

Materials and methods

Based on an optimized cycle recipe for 50 mg/mL sucrose, shelf temperature and pressure settings were varied to simulate potential worst-case conditions during a production scale run. T_p and R_p data were collected during the laboratory experiments using an auto-MTM procedure and correlated with the product's appearance. The sucrose purchased from Sigma-Aldrich (Munich) was of analytical grade and used as received. Solutions of 50, 100, and 200 mg/mL were prepared with double-distilled water from an all-glass apparatus.

Freeze-drying procedure. Freeze-drying was performed with a "FTS Lyostar II" laboratory-scale freeze dryer with the latest version of SMART freeze-dryer software, which also collects MTM data for a user-predefined recipe (Auto-MTM). Three milliliters of solution were filled into 112 20-mL serum tubing vials (Wheaton, Millville, NY) (vial inner area: 5.74 cm²). Vials were placed hexagonally on the middle shelf of the freeze dryer, using one row of empty vials for shielding. Aluminum foil was placed adjacent to the front door inside the chamber to minimize radiation effects. The first cycle used SMART software to obtain a cycle recipe for the product and experimental conditions (T_c= –32 °C, safety margin: 2 °C). The cycle recipe obtained from the SMART run was adjusted to facilitate scale-up conditions (constant shelf temperature over time profile during primary drying). Subsequent runs were performed based on this improvement and then modified for shelf temperature (T_s ± 5 °C), chamber pressure (P_c: 62–200 mTorr), and solute concentration (50–200 mg/mL). T _p and R _p data were recorded and analyzed to describe the influence of the process variations and total solid content on the product morphology.

Freeze-dry microscopy (FDM). Collapse temperatures (T_c) were determined using a "FDCS 196" freeze-drying stage (Linkam Scientific Instruments, Surrey, UK) and an "Axio Imager" microscope (Zeiss, Göttingen, Germany). Images were captured with a 1.3 MPix digital camera and analyzed using "LinkSys" (Linkam) software. The protocol applied was a standard procedure reported in the literature (7, 8). Onset of collapse for the sucrose was determined as follows:

50 mg/mL: T_c = –34°C

100 mg/mL: T_c = –32 °C

200 mg/mL: T_c = –32 °C.

Scanning electron microscopy (SEM). Lyophilized samples were broken into pieces, fixed on aluminum stubs and then carefully gold-sputtered at 20 mA/5 kV (Hummer JR Technics) for about 30 s. Cake morphology was then examined using an Amray 1810 T scanning electron microscope (SEMTech Solutions, CITY, MA) at 20 kV.

Results and discussion

Cycle design by manometric temperature measurement. Figure 1 illustrates the cycle recipe obtained for 50 mg/mL sucrose. The recipe is optimized for the type of excipient, individual T_c of the product, and container system. Note that T_s was automatically lowered by the software after about 6 h to account for a continuous increase in T_p toward the end of primary drying, caused by an increase in R_p (see Figure 4). Product temperature at the sublimation interface did not exceed the collapse temperature during primary drying (see Figure 3 and Table I). A freeze-drying cycle may be denoted as "optimized" when T_p is maintained just below T_c (i.e., 2–3 °C) during primary drying to avoid structural loss (9). For the SMART cycle, total primary drying time during this run was 1136 min and the cycle recipe in primary drying as follows: –39 °C (60 min), –18 °C (358 min), –23 °C (718 min). The endpoint of primary drying was determined using MTM and based on the total difference between the calculated P_ice value by MTM (nonlinear regression analysis of the pressure rise data) and the chamber pressure, P_c. A representative determinant of the endpoint of primary drying is crucial for testing robustness as described in this article. Once all ice has been removed, the pressure rise in the chamber during an MTM measurement is based on leaks and additional heat transfer to the product. Thus, the "fitted" vapor pressure of ice obtained from the MTM procedure equals that of the chamber pressure (see Figure 1) (3–5).

Figure 1

The "adjustment" of the profile showing T _s over time (averaging of the SMART shelf-temperature settings during primary drying to obtain a constant shelf temperature over time) and P _c setting (increase of P _c from 52 to 62 mTorr) was conducted to further improve the recipe and to facilitate scale-up (see Figure 2) (10). T _p and R _p profiles over time for this modified cycle were found in very good agreement to the initial SMART cycle with a maximum temperature of –34.8 °C during the run (see Figure 3 and Table I). Total primary drying time was 1190 min, or about 5% longer than the initial cycle recipe. It is important to note that the product processed in both cycles showed no indication of shrinkage or collapse (see Figure 5a).

Figure 2

Impact of T_sand P_c variation on product temperature. For 50 mg/mL sucrose, variation of shelf temperature of ± 5 °C had very little influence on the profile of T _p over time (see Figure 3). However, an increase in P _c from 52 to 200 mTorr led to a significant increase in T _p (>3 °C), thereby exceeding T _c . In contrast, an increase in solid content to 200 mg/mL and use of identical process conditions revealed only an increase of 1 °C in T _p .

Figure 3

Theoretical modeling of the freeze-drying cycle is an alternative method to gain a better understanding of the impact of process variables on the product-temperature profile. A one-dimensional steady state of the freeze-drying process was used to simulate the influence of T _s and P _c variations on T _p for the 50 mg/mL sucrose solution. Note that T _s and P _c parameters were used according to the experimental design described above (see Figure 3 and Table I). During primary drying, the system is in a "pseudo steady state" and the following equation can be applied to predict product temperature (3, 11):

where ΔH _s is the heat of sublimation of ice (670 cal/g used as constant), P_ice is the vapor pressure of ice (Torr, known from the first experiment), P _c is the chamber pressure (Torr), R _p and Rs are the product and stopper resistance, respectively (cm² ×Torr ×h/g, known from the first experiment), A_v is the vial outer cross-sectional area (6.83 cm²), K_v is the vial heat transfer coefficient (10⁴ ×cal/sec ×cm² ×K, cf.) (see Table I), T _s is the shelf-inlet temperature (°C) and T_b the temperature at the bottom of the vial (°C).

Table I illustrates the theoretical values obtained from the steady-state model for the maximum product temperature (T_p-max) when using a ± 5 °C shelf temperature variation at constant pressure (62 mTorr) or an increase in chamber pressure from 62 to 200 mTorr at constant T _s .

Table I: Summary of process conditions for 50–200 mg/mL sucrose solutions and corresponding maximum product temperature data (Tp-max) during primary drying.

The modeled results obtained for T_p-max were found in fairly good agreement to the experimental data. Note that this observation has been reported in the literature (11). However, the effect of T_s and P _c on product temperature was lower in the experiments performed relative to the theoretical calculations, in particular for applied pressure changes. Although such simulation is useful for scale-up, rational design of the freeze-drying process and troubleshooting, simulations cannot eliminate the need for careful experimentation (12).

Figure 4

Product resistance. R_p data determined for products dried by the optimized cycle increased almost linearly over L_dry, indicating an absence of shrinkage and microcollapse (see Figure 4) (4, 5). The variation of T _s by ± 5°C showed no clear impact on R_p. However, the pressure increase up to 200 mTorr resulted in a significant decrease in R _p over L_dry. Microcollapse as well as shrinkage in these products could be detected visually and by SEM (see Figure 5b).

Figure 5

Note that an increase in solute concentration led to a similar R _p curve shape relative to the samples at high P_c, but the product matrix revealed much greater robustness to elevated temperature (see Figures 4 and 5c). This observation was in excellent agreement with the T _c difference measured by FDM between 50 mg/mL (T_c= –34° C) and 200 mg/mL (T _c = –32 °C) sucrose solutions. Note that the dependency of sucrose concentration on T _c was recently investigated but is in contrast to data obtained in early studies on a different amorphous system (8, 13).

Conclusion

The study indicates that safety margins for shelf temperature and chamber pressure for a given product can be pre-evaluated in a laboratory using advanced PAT tools and therefore may allow for a rational QbD approach in the future. In addition, product resistance data obtained from MTM may provide valuable information about microcollapse in the product ("benchmark" properties). This information can be very useful in interpreting the effect of temperature or pressure deviations on a product.

Henning Gieseler, PhD,* and Stefan Schneid work in the Division of Pharmaceutics at the University of Erlangen, Erlangen, Germany 91058, tel. +49 9131 85 29545, gieseler@freeze-drying.euTony Kramer works at the Ecolab Research Center in Eagan, Michigan.

*To whom all correspondence should be addressed. Submitted: Jan. 23, 2008. Accepted. Apr. 23, 2008.

What would you do differently? Email your thoughts about this paper to ptweb@advanstar.com and we may post them to the site.

References

1. J. Wechsler, "Modernizing Pharmaceutical Manufacturing," Pharm. Technol. 26 (2002).

2. FDA/ORA, "Compliance Policy Guide, Sub Chapter 490.100," Process Validation Requirements for Drug Products and Active Pharmaceutical Ingredients Subject to Pre-Market Approval (CPG 7132c.08).

3. M.J. Pikal and H.R. Costatino, "Lyophilization of Biopharmaceuticals" in Biotechnology: Pharmaceutical Aspects, R.T. Borchard and C. R. Middaught, Eds. (AAPS Press, Arlington, VA, Vol. II, 2004).

4. X. Tang, S.L. Nail, and M.J. Pikal, "Freeze-Drying Process Design by Manometric Temperature Measurement: Design of a Smart Freeze-Dryer," Pharm. Res. 22 (4) 685–700 (2005).

5. H. Gieseler H, T. Kramer, and M.J. Pikal, "Use of Manometric Temperature Measurement (MTM) and SMART Freeze Dryer Technology for Development of an Optimized Freeze-Drying Cycle," J. Pharm. Sci.96 (12), 3402–3418 (2007).

6. H. Gieseler, "Process Analytical Technology for Freeze-Drying: Cycle Optimization in the Laboratory," Eur. Pharm. Rev., 1, 2007.

7. F. Fonseca et al., "Collapse Temperature of Freeze-Dried Lactobacillus bulgaricus Suspensions and Protective Media," Biotechnol. Prog. 20, 229-238 (2004).

8. E. Meister and H. Gieseler, "Evaluation of Collapse Temperatures by Freeze-Dry Microscopy: Impact of Excipient Concentration on Measured Transition and the Overall Dependence on Measurement Methodology," in Proceedings of Fifth World Meeting on Pharmaceutics and Pharmaceutical Technology (Geneva, Switzerland, March 2006).

9. S. Rambhatla et al., "Cake Shrinkage During Freeze Drying: A Combined Experimental and Theoretical Study," Pharm. Dev. Technol, 1, 33–40 (2005).

10. S. Tchessalov, "Principles of Lyophilization Cycle Scale Up" in Processing CPPR Freeze Drying of Pharmaceuticals and Biologicals Conference (Garmisch-Patenkirchen, October 2006).

11. M.J. Pikal, "Use of Laboratory Data in Freeze Drying Process Design: Heat and Mass Transfer Coefficients and the Computer Simulation of Freeze Drying," J. Parent. Sci. Technol. 39 (3), 115–138 (1985).

12. J.P Dolan, "Use of Volumetric Heating to Improve Heat Transfer During Vial Freeze Drying" in Mechanical Engineering (PhD Thesis, Virginia Polytechnic Institute, Blacksburg, VA, 1998).

13. M.J. Pikal and S. Shah, "The Collapse Temperature in Freeze Drying: Dependence on Measurement Methodology and Rate of Water Removal from the Glassy Phase," Int. J. Pharm. 62, 165–186 (1990).