I recently had the privilege of chairing the 3rd Annual Lyophilization Conference held February 25-26 in Boston, MA. The conference was well organized and offered thoughtful insights for attendees who were relatively new to the field of lyophilization as well as to those who had been working in the field for decades. This article is a brief summary of the conference.
The speakers were encouraged to consider their talks from the perspective of either cost reduction or enabling technologies. Cost reduction is becoming an increasingly important aspect of lyophilization processes as the pressure steadily increases to reduce the cost of manufacturing of an existing product in addition to reduce the costs of running a lyophilization facility producing a multitude of products. The audience was treated to lectures from pharmaceutical scientists and vendors on topics such as excipient selection, lyophilization cycle design, container-closure considerations, scale-up for clinical and commercial production, process robustness and facility utilization. There were also several talks on alternate approaches to drying such as spray-drying, bubble-drying and the preservation of mammalian cells. These talks, as well as an appreciation for the continued challenges within lyophilization science, made for a thought provoking meeting.
Several speakers, including Willow DiLuzio of Millennium, discussed excipient selection as an initial step in the development of a lyophilized product. In brief, we should assume that a lyophilized powder should be stable for at least 2 years, reconstitute within 2-5 minutes and be relatively easy to manufacture. A specific upside would be room temperature storage: there are at least three proteinbased products in which ambient storage enables their application in their specific therapeutic area. All agreed that there is a strong desire to utilize globally acceptable excipients that reduce the need to demonstrate safety and suitability of a novel excipient or one that has not been previously utilized via a certain route of administration. It was agreed that not all buffers, for example, are suitable.
Acetate will volatize and sodium phosphate will selectively crystallize resulting in dramatic pH shifts upon freezing. In fact, there is a fairly narrow choice of excipients to select from whether one is searching for a buffer (Tris, succinate, histidine, glutamate), a cryo-protectant (sucrose, trehalose), surfactants such as polysorbates 20 and 80, and bulking agents such as mannitol and glycine. Other excipients can be explored, but they may have little advantage over these widely used excipients. Finally, care must be taken when selecting excipients in case there are specific concerns with oxidation or metal catalyzed degradation. Not only are these excipients used to stabilize the protein in solution and in the dried state, but they must also have properties that enable freeze-drying in a timely manner. While sucrose is a well established lyo-protectant, it does exhibit a low glass transition temperature. This can result in slower lyophilization cycles. The combination of an appropriate level of sucrose with a suitable bulking agent (depending on the concentration of the active ingredient) and a lyophilization cycle which maintains the proper product temperature during primary drying, can yield a product that is not only stable but also survives the rigors of the freezedrying process itself. This last point was elegantly made by Alexander Klibanov of MIT. Professor Klibanov discussed the use of FTIR to monitor the reversible denaturation of proteins during the freeze-drying process. This reversible denaturation can often be a precursor to further chemical modification of a protein (intermolecular disulfide bonding, for example), but can also be managed by proper formulation minimizing the level of denaturation in the first place.
Excipients should be selected and balanced to provide characteristics that permit freeze-drying with product temperatures just below the critical process parameters. The use of arginine, a well -regarded stabilizer from the perspective of liquid protein stability, is difficult to process due to its low glass transition temperature of -41°C. Again, the cycle and the formulation are intimately linked. David Hamilton of Merck grounded the audience with a thorough discussion of the fundamentals of a good freeze-drying cycle and focused on the role of nucleation. Ice nucleation is often very inconsistent, vial-to-vial, in the early stages of product cooling. This variability results in different thermal properties across the freeze-dryer resulting in variable drying rates. The use of a thermal treatment step, or annealing, after the initial freeze serves to erase this individual thermal history and provide a better opportunity for uniform lyophilization (excluding the variability introduced by the mechanics of the freeze-dryer itself). Similarly, Suman Luthra of Pfizer presented the application of annealing of the freeze-dried powder itself. By annealing dried powders at temperatures greater than room temperature but below the glass transition temperature of the dried powder, she argued that annealing results in a deliberate structural relaxation, reduced global mobility, and reduced energy within the powder resulting in enhanced stability of several small molecules as well as an IgG. These talks highlighted that a lyophilization cycle not only provides a means to a freeze-dried powder, but, if optimized, can provide characteristics that enhance the pharmaceutical properties of the product. While liquid products are often most desirable from a marketing perspective, via their ease of administration, this consideration is less critical in emerging markets. In those regions in which cold chain facilities are less widely available, and where drug availability often drives market preference, the ability to provide a lyophilized powder stable for several years at room temperature could not only be a competitive advantage but could also open new markets to existing products. A thoughtful selection of excipients and freeze-drying cycle can enable this type of opportunity.
In terms of product definition, we must also consider container closure systems. There were three talks specifically dedicated to container design, selection and the ability to monitor container integrity during the inspection process. Vial selection and some of the problems faced during lyophilization such as “pop-ups” in which the stopper pops up after the shelves have been decompressed were discussed. Options to manage this problem include proper vial neck design during stopper/vial selection as well as the use of coated stoppers. Vial breakage can be managed by either changing the formulation (reducing the level of mannitol for example) and/or fill volume (reducing to approximately half the vial volume). One possible solution for vial breakage is to utilize a more robust, molded vial designed specifically for lyophilization applications. Ensuring that vials, or batches, are not lost due to vial breakage can help to ensure a smooth flow of product to the market.
Stopper selection and crimping were discussed. Specifically, we must pay attention not only to the relationship between the vial and stopper, but to the stopper and seal during the shelf compression process. Recent regulatory trends, especially in Europe, are migrating toward verifiable assurance of container-closure integrity of each vial (see EudraLex, Volume 4, Annex 1 published 14 February 2008). One of the drivers for this concern is the occasional practice of crimping sealed vials in areas quite distant from the lyophilization room. One possible solution is to utilize novel seals that enable crimping within the lyophilizer itself, avoiding the need for external capping and reducing the need to build facilities in which cappers must operate within class A space utilizing sterilized crimps. Ensuring container closure integrity, via a 100% automated test, was also discussed. The use of head space analyzers during the inspection or packaging process is being developed to accommodate high speed testing with minimal false rejects. In addition, there is ongoing development of automated NIR analyzers to monitor moisture levels within the vials during inspection. While perhaps not as sensitive as Karl Fisher analysis, this approach, if married to headspace analyses, could provide additional assurance of container integrity as well as provide data across the batch, during validation or characterization of a chamber.
Issues of scale-up and tech transfer were touched upon by several speakers. First, there was broad agreement that scale-up and transfer between freeze-dryers needs to focus on product temperature and the pressure difference at the sublimation surface within the vial. These are critical dependent variables. Other considerations (independent variables such as shelf temperature, chamber pressure) are critical in that they control the rate of freezing and sublimation and thus affect the structure of the matrix, product temperature and pressure drop.
Understanding the critical process parameters is key to process design. As we look forward to scale-up, we also need to thoughtfully consider the impact of the specific equipment that will be used. While it would be ideal to perform development work at commercial scale, this is prohibited by time, material availability and cost. One solution to this dilemma is to utilize small scale freeze-dryers (0.1 to 3 m2) which are designed with similar scaled down characteristics to clinical or pilot scale dryers (up to 8 m2) and ultimately commercial scale dryers up to 55 m2. In this manner, the relationship between the lyophilizer chamber and condenser, the flow of the heat transfer fluid through the shelves, the diameter of the duct connecting the lyophilization chamber to the condenser (which can choke the flow of water vapor), as well as such details like valve design, process control instrumentation and door design are harmonized across scales. While this may be ideal, it can often be difficult to achieve when one has to work with freeze-dryers at a contractor, at various sites within a company’s network of facilities as well as the simple variability that occurs across freeze-dryers of the same model.
An alternate approach is to perform a thorough characterization of lab and commercial scale freeze-dryers and their performance relative to minimum controllable pressure versus sublimation rate. Serguei Tchessalov of Pfizer presented several comparisons between small scale lab freeze-dryers and commercial scale dryers (0.42 m2 versus 42 m2), in which he was able to mimic commercial conditions within a small scale dryer. This “scale-down” approach provides useful data when seeking to ensure that the scale-up will be successful and robust. As mentioned before, the main objective of this comparison is to understand equipment capabilities so that the critical process parameters such as product temperature and pressure differential can be maintained across freeze-dryers. Other properties of the lyophilizer performance that should be assessed include the influence of heat transfer fluid temperature gradients between the inlet and shelf surface and the affect these can have on freezing rates. Commercial scale freeze dryers often have greater cooling capacity than smaller lab scale dryers, therefore they may need to be run at lower freezing rates (often 0.5 °C/min) in order to mimic small scale as well as ensure uniform freezing across the shelf. Heat transfer coefficient, as a function of chamber pressure, should be determined for the various freeze-dryers. Understanding these differences enables one to model the steady-state sublimation process during primary drying and predict at any scale when primary drying will be complete. This modeling approach also enables understanding the effect of process deviation in chamber pressure or shelf temperature, and the impact this deviation may have on product temperature and sublimation rate.
One can also model the characteristics of a freeze-dryer. One of our speakers showed data from computer modeling of the temperature and pressure gradients that exist within a freeze dryer, on the basis of its design and performance. He assessed the impact of heat transfer fluid flow and door design on shelf temperature, the location of the duct to the condenser and its effect on pressure within the drying chamber as well as nitrogen distribution throughout the chamber which is often thought to be uniform but can set up a gradient within the chamber. Having a thorough understanding of the product’s critical process attributes marries well to lyophilization cycle design which then leads to robust scale-up. The successful scale-up is then dependent on not only understanding the lyophilization cycle but also the equipment which is to be utilized at both lab and commercial scale.
Several speakers touched on the subject of robustness and shared a wide variety of experiences. David Hamilton described his view of performing several robustness cycles at commercial scale, with full characterization of the chamber (moisture mapping), as well as release and stability testing. These studies would be designed to assess the process capability within the commercial dryers and would define the design space in which the cycle and dryer can operate. The approach is one in which the cycle is not pushed to failure. He advised the group to consider a control cycle followed by 4 cycles to assess the effect of elevated shelf temperature and chamber pressure as well as reduced shelf temperature and chamber pressure as separate variables. If one wished to be more aggressive, one could combine these into 2 cycles by combining elevated shelf temperature and chamber pressure and combining reduced shelf temperature and chamber pressure. Note that this may require either a trial run or the use of a pressure rise test during the conservative cycle to ensure the completeness of primary drying. While this type of data would provide a high degree of confidence in the acceptability of materials produced by such a deviation, it does consume several slots within a production freeze-dryer. This would have to be negotiated as part of the process validation and scale-up.
Serguei Tchessalov described an alternate approach utilizing a lab scale dryer and assessing various process steps either individually or in sets. Variables to consider include freezing (ramp rates, annealing temperature, step duration), primary drying (set pressure ± 20mT, shelf temperature ± 5°C, step duration) and secondary drying (pressure, shelf temperature, step duration). One needs to consider the benefit of these datasets when defining design space as well as when having discussions with a quality group or regulatory agency as to the suitability of release of a batch after a deviation has occurred.
A third approach takes advantage of a Design of Experiment (DOE) method. While well established for many years, the application of DOE experiments to lyophilization processes has attracted more attention due to an increase in process analytics technologies (PAT) and recent emphasis by the regulatory authorities on Quality by Design (QbD) approaches. Several software options exist that aid in the analysis of these experiments, but the results are only as good as the design of the experiments and the collection of the data. Prior to performing experiments, a risk analysis and failure mode and effects analysis (FMEA) will help to ensure that the correct variables are assessed. One should also consider the need to assess and analyze multiple variables simultaneously. While these approaches have been used broadly for many years, there is still considerable work associated with convincing quality and regulatory staff of their benefits.
One aspect of performance, often observed during robustness studies, is the question of cake defects. Several speakers presented a wide variety of defects including meltback, shrinkage, collapse and so on. While there was rough agreement on how best to describe these defects, linking their appearance to cause remains a substantial task and should be discussed further, perhaps to the point of publishing guidance on how best to describe cake defects.
Once the process has been scaled, validated and in production, there is still the task of monitoring ongoing production. Often, the processes that are implemented are not optimized and may have robustness problems resulting in an elevated number of rejected vials, lot batches, or variable vial-to-vial levels of moisture which can impact stability. One speaker presented an elegant argument in which he contrasted the costs associated with investing in additional development work to optimize cycles versus the costs associated with lost vials or batches as well as poor utilization of facility time. While there is no justification for constant tinkering of a production cycle, it was agreed that having discrete goals associated with increasing the robustness of a cycle (and vial yield) as well as optimizing the cycle duration, enabling more batches to go through a facility in a year, was warranted. A thoughtful discussion between development scientists and manufacturing colleagues may result in the funding of development studies and production runs that could save tremendous time and money over a 6-24 month period. Ultimately, the proper balance of cost challenges (resources, materials, opportunity cost) and quality challenges (risk knowledge and management, increased yields, no rejects or deviations…) should result in the lowest cost per vial.
The meeting would not be complete if we did not have a thoughtful discussion of alternate approaches to drying. Spray-drying, in particular, was discussed for the production of dry powders suitable not only for pulmonary administration but also for rapid reconstitution within a medical device. One concept is the generation of a small dual chambered pouch containing spray dried powder in one chamber, diluent for reconstitution in the other, and the ability to conveniently combine the two followed by injection of a vaccine through a fine needle. This is an enabling approach for the delivery of vaccines that traditional vial lyophilization struggle to achieve. In another example, bubble drying by the use of supercritical fluids could enable the production of dried particles of live virus vaccines of a uniform particle size. These particles, if stable at room temperature, could enable the development of vaccines for emerging markets in which a cold chain often is not available. Finally, foam drying was discussed not only in the context of traditional pharmaceuticals but also for the application of the conservation of mammalian cells. In this specific application the stress of traditional lyophilization and spray drying, may be prohibitive to cell viability. Foam drying, in contrast, may be suitably gentle as to enable the drying and room temperature storage of mammalian cells for a variety of treatments.
In brief, this conference touched on many of the elements of the design and implementation of a lyophilization process for the stabilization of labile compounds. Topics included excipient selection, rational cycle development, container-closure problems and solutions, scale-up, robustness and commercial scale optimization. In addition, we discussed alternate technologies which may enable products to be developed that cannot currently be developed using traditional lyophilization technologies.
Dr. Warne is the Senior Director of the Formulations Group at Pfizer BioTherapeutics R&D in Andover, Massachusetts. Nick has been at Pfizer, formerly Wyeth BioPharma and Genetics Institute, for 20 years and has focused on protein stabilization, formulation development and drug product process development. Nick holds numerous protein formulation patents and, with his group, has made over 100 presentations at national meetings and in journals.